Please note: This file provides a complete, temporally ordered log of changes that went into every version of Perl. If you'd like more detailed information, please consult the comments in the individual patches posted to the perl5-porters mailing list. Patches for each individual change may also be obtained through ftp and rsync--see pod/perlhack.pod for the details. For information on what's new in this release, see pod/perldelta.pod. [The "CAST AND CREW" list has been moved to AUTHORS.] NOTE: Each change entry shows the change number; who checked it into the repository; when; description of the change; which branch the change happened in; and the affected files. The file lists have a short symbolic indicator: ! modified + added - deleted +> branched (from elsewhere) !> merged changes (from elsewhere) The Message-Ids in the change entries refer to the email messages sent to the perl5-porters mailing list. You can retrieve the messages for example from http://www.xray.mpe.mpg.de/mailing-lists/perl5-porters/ -------------- Version v5.8.9 Maintenance release working toward v5.8.9 -------------- ____________________________________________________________________________ [ 30141] By: nicholas on 2007/02/05 22:46:22 Log: Integrate: [ 29520] Rename FBM_TABLE_OFFSET to PERL_FBM_TABLE_OFFSET prior to moving it to a header. [ 29522] Convert the -1 to a more explicit PERL_FBM_FLAGS_OFFSET_FROM_TABLE. [ 29524] Move PERL_FBM_TABLE_OFFSET and PERL_FBM_FLAGS_OFFSET_FROM_TABLE to sv.h [but not this bit: Stow BmRARE in the SvPVX, and so delete xbm_rare. Can you see what it is yet? ] [ 29525] Add BmPREVIOUS_set(), and use it in the one place that BmPREVIOUS is modified. [ 29538] Correct some assumptions about PVBM table offset in B.xs [ 29541] In theory Perl_magic_setbm() should clear SvTAIL() too. In practice, I don't think that it matters as this routine is never actualy called because nothing exposes PVBMs to the world so that the world can assign to them. [ 29635] Eliminate BmPREVIOUS_set - with the complexity gone from how the datum is stored, there's no need for it. [ 29637] In Perl_fbm_compile, really rarest should be U32 not I32, as it is set from a U32, and used as an array index. Branch: maint-5.8/perl ! sv.h !> ext/B/B.xs mg.c util.c ____________________________________________________________________________ [ 30140] By: nicholas on 2007/02/05 22:01:07 Log: Integrate: [ 27195] Subject: [PATCH] Make XCOPY not prompt every two seconds when rebuilding on Win32 From: demerphq Date: Mon, 13 Feb 2006 09:41:37 +0100 Message-ID: <9b18b3110602130041q2d64122am5c83392dd75e0413@mail.gmail.com> [ 27200] Subject: [PATCH] stat() on Windows doesn't handle trailing slashes/backslashes correctly From: "Jan Dubois" Date: Wed, 15 Feb 2006 17:19:29 -0800 Message-ID: <017901c63297$08e1e3f0$2217a8c0@candy> [ 27283] Subject: [PATCH] Implement ${^WIN32_SLOPPY_STAT} From: "Jan Dubois" Date: Tue, 21 Feb 2006 12:36:31 -0800 Message-ID: <019601c63726$7fcca200$6062a8c0@candy> [ 27421] Get rid of unnecessary "rmdir /s" calls in Win32 makefiles They only existed for the benefit of old shells that didn't understand "rmdir /s /q", but since change #27195 effectively drops support for old shells that don't understand "xcopy /y" it seems unnecessary now. Windows NT4, 2000, XP and onwards all support "rmdir /s /q", so it's probably only the Windows 9x shells that lose out. [ 27500] The first step in supporting VC++ 8.0. Added macros for the free .NET 2.0 SDK and Visual Studio 2005. Added these macros to "if" conditions that were appropriate. Added a fix to prevent the .NET 1.1 and 2.0 SDK compilers from attempting to build perlglob.exe (they are both missing the setargv.obj file necessary to do it). [ 27503] Additions to README.win32 and win32/Makefile based on comments from Steve Hay. [ 27528] Subject: [PATCH] win32_pclose() error exit doesn't unlock mutex From: Jan Dubois Date: Thu, 16 Mar 2006 16:30:23 -0800 Message-ID: <02c701c64959$fac63ad0$6062a8c0@candy> [ 27530] Install POD files into "lib\pods" rather than "lib\pod" on Win32 to avoid clashing with "lib\Pod" (cf. perl change #26363) Also remove a couple of unused macros from the makefiles [ 27532] Make "distclean" target on Win32 clean up temporary HTML directory that gets left behind after running the "install" target. Also clean up a couple of leftover pod2html cache files. [ 27555] Merge some changes from ActivePerl 5.8.8.816 See: http://downloads.activestate.com/ActivePerl/src/5.8/AP816_diff.txt These changes provide better support for building XS extension modules using VC++ or MinGW when perl itself was built with the opposite of those two compilers. [ 27556] Merge some more changes from ActivePerl 5.8.8.816 See (again): http://downloads.activestate.com/ActivePerl/src/5.8/AP816_diff.txt These changes improve support for the MS 64-bit compiler in win32/Makefile (and tentatively make the same changes to win32/makefile.mk), and also switches the now deprecated -Gf switch to -GF for all flavours of VC++ instead of just the latest ones. (See: http://www.xray.mpe.mpg.de/mailing-lists/perl5-porters/2006-02/msg00353.html for a comment on the s/-Gf/-GF/ change.) [ 27888] Subject: [PATCH] win32/Makefile changes for Windows 64 bit From: "Jan Dubois" Date: Mon, 17 Apr 2006 01:48:34 -0700 Message-ID: <036401c661fb$bc7a1200$2217a8c0@candy> and make similar changes to win32/makefile.mk and delete a spurious extra line from that file Branch: maint-5.8/perl !> README.win32 installperl win32/Makefile win32/bin/pl2bat.pl !> win32/config_H.gc win32/config_H.vc win32/makefile.mk !> win32/win32.c win32/win32io.c wince/wince.c ____________________________________________________________________________ [ 30139] By: nicholas on 2007/02/05 21:19:54 Log: Integrate: [ 27180] Subject: [Patch] Enhance Hash::Util From: demerphq Date: Mon, 13 Feb 2006 11:39:33 +0100 Message-ID: <9b18b3110602130239w311d05fcr776ae8333776ca2e@mail.gmail.com> [ 27181] Hash::Util tests should check if Hash::Util has been built, not List::Util (spotted by Rafael). [ 27257] Add tests for the previously untested Hash::Util::all_keys(). Branch: maint-5.8/perl +> ext/Hash/Util/Changes ext/Hash/Util/Makefile.PL +> ext/Hash/Util/Util.xs ext/Hash/Util/lib/Hash/Util.pm +> ext/Hash/Util/t/Util.t - lib/Hash/Util.pm lib/Hash/Util.t !> MANIFEST win32/Makefile win32/makefile.mk ____________________________________________________________________________ [ 30138] By: nicholas on 2007/02/05 19:37:53 Log: Integrate: [ 25288] Make change 25286 better Thanks to John E. Malmberg for pointing this out. [ 26041] Subject: [PATCH] Remove unused USING_WIDE code from win32 and wince branches From: "Jan Dubois" Date: Mon, 7 Nov 2005 20:39:02 -0800 Message-ID: <078001c5e41e$585c4140$6401a8c0@candy> [ 26044] Remove an unused variable left over from change #26041 Branch: maint-5.8/perl !> win32/Makefile win32/bin/pl2bat.pl win32/ext/Win32/Win32.xs !> win32/makefile.mk win32/perlhost.h win32/win32.c win32/win32.h !> win32/win32io.c wince/perlhost.h wince/win32.h wince/wince.c ____________________________________________________________________________ [ 30137] By: nicholas on 2007/02/05 19:11:17 Log: Integrate: [ 29823] Various cleanup and factorization by Schwern for tests for warnings and features [ 29824] Simplify the ok/not ok logic (another refactorisation by Schwern) [ 29825] Use the $Is_MacOS variable [ 30123] Enhance t/lib/common.pl to provide conditional TODOs using eval'd code. [ 30136] t/lib/common.pl can be warnings clean. Branch: maint-5.8/perl !> t/lib/common.pl ____________________________________________________________________________ [ 30135] By: nicholas on 2007/02/05 18:48:59 Log: Integrate: [ 26473] ExtUtils::ParseXS needs to stack $filepathname to make #line directives in #INCLUDEs work. [ 28856] Upgrade to ExtUtils-ParseXS-2.16. There actually are a couple of changes in bleadperl not in the new version, so version was bumped to 2.16_01. [ 29269] Subject: [PATCH] ParseXS.pm: small optimization for "Usage: ..." constant strings From: Alexey Tourbin Date: Tue, 14 Nov 2006 05:45:40 +0300 Message-ID: <20061114024540.GA21986@localhost.localdomain> [ 29418] Upgrade to ExtUtils-ParseXS-2.17. Actuall just caught up with most of the patches to bleadperl. The remaining change missing is #29252, so, the version has been bumped to 2.17_01. [ 30134] Upgrade to ExtUtils-ParseXS-2.18 Branch: maint-5.8/perl !> lib/ExtUtils/ParseXS.pm ____________________________________________________________________________ [ 30133] By: nicholas on 2007/02/05 18:05:43 Log: Integrate: [ 26174] Upgrade to PathTools-3.14 [ 26318] Upgrade to PathTools-3.14_01 [ 26319] Re-instate the old perl getcwd as _perl_getcwd, and use it if loading the XS fails. (Needed to bootstrap the core build) [ 26508] Upgrade to PathTools-3.15 [ 27398] Upgrade to PathTools-3.17 [ 28001] Upgrade to Pathtools-3.18 [ 28004] Added Changes file left out from change #28001. [ 28511] Subject: [PATCH] optimisation for File::Spec::abs2rel where both args are relative From: Brendan O'Dea Date: Thu, 6 Jul 2006 01:02:41 +1000 Message-ID: <20060705150241.GA2244@londo.c47.org> [ 28551] Upgrade to PathTools 3.19 [ 28948] Upgrade to PathTools-3.21 [ 28983] Update to PathTools-3.22. [ 29004] Upgrade to PathTools-3.23. [ 29356] Upgrade to PathTools-3.24. Branch: maint-5.8/perl +> lib/File/Spec/t/tmpdir.t !> MANIFEST ext/Cwd/Changes ext/Cwd/Cwd.xs ext/Cwd/t/cwd.t !> lib/Cwd.pm lib/File/Spec.pm lib/File/Spec/Cygwin.pm !> lib/File/Spec/Mac.pm lib/File/Spec/Unix.pm !> lib/File/Spec/VMS.pm lib/File/Spec/Win32.pm !> lib/File/Spec/t/Spec.t lib/File/Spec/t/crossplatform.t ____________________________________________________________________________ [ 30132] By: nicholas on 2007/02/05 17:37:09 Log: Integrate: [ 26071] Subject: [PATCH @26062] wince patch From: "Konovalov, Vadim" Date: Wed, 9 Nov 2005 20:22:52 +0300 Message-ID: <7DD1BE2C50259746ABB8683672D2089E12DC02@itotest-1.spb.lucent.com> [ 26072] Really apply change #26071. [ 26140] Subject: [PATCH @26137] for wince From: "Konovalov, Vadim" Date: Wed, 16 Nov 2005 20:57:13 +0300 Message-ID: <7DD1BE2C50259746ABB8683672D2089E12DC9F@itotest-1.spb.lucent.com> [ 26200] Subject: [PATCH @26198] perlce touches From: Konovalov, Vadim Date: Nov 23, 2005 8:55 PM Message-ID: <7DD1BE2C50259746ABB8683672D2089E12DDC3@itotest-1.spb.lucent.com> [ 26201] Delete a README made redundant by change 26200. Branch: maint-5.8/perl - wince/README.perlce !> INSTALL MANIFEST README.ce configpm ext/Cwd/Cwd.xs makedef.pl !> wince/Makefile.ce wince/config.ce wince/config_H.ce !> wince/config_h.PL wince/config_sh.PL wince/wince.c ____________________________________________________________________________ [ 30131] By: nicholas on 2007/02/05 17:13:40 Log: Integrate: [ 27202] Upgrade to CGI-3.16, with version bump on CGI.pm for documentation fixes not yet integrated. [ 27255] Subject: Patches: B, CGI, ExtUtils::MM_Unix From: "Joshua ben Jore" Date: Sun, 19 Feb 2006 02:58:10 -0600 Message-ID: [ 27354] Upgrade to CGI.pm-3.17, but continuing the version bump for unintegrated changes. [ 27873] Upgrade to CGI.pm-3.19. [ 28082] Upgrade to CGI-3.20 [ 28732] Fix to problem where CGI can lose the filehandle during an upload. Patch posted to the Debian bug list by Eric Wong . [ 28746] Upgrade to CGI.pm-3.21 [ 28752] Upgrade to CGI.pm-3.22. [ 28930] Upgrade to CGI.pm-3.25 Branch: maint-5.8/perl !> AUTHORS lib/CGI.pm lib/CGI/Carp.pm lib/CGI/Changes !> lib/CGI/Cookie.pm lib/CGI/Fast.pm lib/CGI/Util.pm !> lib/CGI/t/cookie.t lib/CGI/t/function.t lib/CGI/t/html.t ____________________________________________________________________________ [ 30130] By: nicholas on 2007/02/05 16:25:02 Log: Integrate: [ 24535] [perl #35878] goto &xsub that croaks corrupts memory When an XS sub is called, a CxSUB context shouldn't be pushed. Make goto &xs_sub mimic this behaviour by first popping the old CxSUB [ 24642] add pointer to email explaining why eval {goto &foo} is banned Branch: maint-5.8/perl !> pp_ctl.c t/op/goto_xs.t ____________________________________________________________________________ [ 30127] By: nicholas on 2007/02/05 11:57:18 Log: Integrate: [ 24361] move the SETJMP exception-handing definitions from scope.h to cop.h so that a JMPENV* entry can be added to struct block_eval [ 24362] Revert change #15705 (Core dump in 'leavetry') This fixes bug #34682, reintroduces bug #8738 (ID 20020301.011), and reintroduces an eval optimisation for innter runops levels [ 24363] Better fix for #8738 (Core dump in 'leavetry') When in an inner runops loop (eg via a tie or sort), an eval needs a new JMPENV pushing by S_docatch. If an exception is raised, control is returned to S_docatch, and it must determine whether the eval that trapped the exception is an inner eval or an outer one. In the former case, restart the loop, in the latter case, rethrow the exception. This is determined by whether we are still at the same PL_curstackinfo level. This fails in the case of SPLICE(), which pushes a new SETJMP and runops level, but not a new stackinfo level. There may be other code which does similar. The solution is to store the current value of PL_top_env in each pushed CxEVAL, and see if it's still the same as PL_top_env when the exception is handled. [ 24387] add test for [perl #34682] leaving eval via last in inner runops Branch: maint-5.8/perl ! cop.h pp_ctl.c !> scope.h t/op/eval.t ____________________________________________________________________________ [ 30122] By: nicholas on 2007/02/04 22:37:19 Log: Clear up a maint specific compiler warning - in 5.8.x we have more than 16 entries in the arena data structures, so SVt_PTE is > 15, so won't fit in 4 bits. It's not actually needed by the sanity test, so change its "type" to 0. Branch: maint-5.8/perl ! sv.c ____________________________________________________________________________ [ 30121] By: nicholas on 2007/02/04 21:32:10 Log: Integrate: [ 23753] Turn INIT_TLS_AND_INTERP into a static function [ 23755] Foolishly I committed change 23753 before remembering to test without ithreads. No tests => bugs. This should fix them. [ 30119] Change 23753 wasn't quite a perfect refactoring, as it omitted calling PERL_SET_THX(my_perl); when !PL_curinterp for non-ithreads. However, this would not have made a difference for any configuration buildable from the blead source alone. Branch: maint-5.8/perl !> perl.c ____________________________________________________________________________ [ 30118] By: nicholas on 2007/02/04 19:27:45 Log: Integrate: [ 23161] make -Dl log runops loop entry and exit, and jump level popping Branch: maint-5.8/perl !> dump.c scope.h ____________________________________________________________________________ [ 30117] By: nicholas on 2007/02/04 19:00:56 Log: local $[ works quite nicely in maint, so un-TODO it. Branch: maint-5.8/perl ! t/op/local.t ____________________________________________________________________________ [ 30116] By: nicholas on 2007/02/04 18:51:01 Log: Put the assignment to $XS_VERSION in a BEGIN block too. Branch: maint-5.8/perl ! ext/DynaLoader/DynaLoader_pm.PL ____________________________________________________________________________ [ 30115] By: nicholas on 2007/02/04 18:49:54 Log: Integrate: [ 24041] Don't use vars in DynaLoader Remove left-over maintenance code [ 24271] Subject: [PATCH] Symbian port of Perl From: Date: Mon, 18 Apr 2005 13:18:30 +0300 Message-ID: [ 27522] Avoid hardcoded $(OBJ_EXT) [ 27549] Move DynaLoader.o into libperl.so. This avoids the need to statically link DynaLoader into the stub perl executable and make libperl.so provide all the code needed to get a functional embedded perl interpreter up running. As a side effect this also moves DynaLoader into libperl.a for non-useshrplib builds. Fixes [perl #32539] [ 27566] Adjust docs to reflect that DynaLoader, as of change 27549, is part of libperl. [ 28226] Subject: [PATCH] Test script for DynaLoader From: Sébastien Aperghis-Tramoni Date: Thu, 18 May 2006 16:46:00 +0200 Message-ID: <1147963560.446c88a8891e1@imp1-g19.free.fr> [ 28227] Nits in the DynaLoader test: make it warning-clean, and avoid checking OS- and locale-specific messages [ 28228] We can now remove DynaLoader from 1_compile.t [ 28274] Better handling for error-checking From: IRC/#p5p/Maddingue (Sebastien Aperghis-Tramoni) [ 28308] Subject: Re: [PATCH] Test script for DynaLoader From: demerphq Date: Thu, 25 May 2006 13:20:53 +0200 Message-ID: <9b18b3110605250420m7431bfbeve44aada2025264f7@mail.gmail.com> [ 28318] use the same arg names everyone else does and handle undef filename [ 28321] Subject: [PATCH] DynaLoader.t: less assumptions From: Jarkko Hietaniemi Message-ID: <44796AF2.8060903@gmail.com> Date: Sun, 28 May 2006 12:18:42 +0300 [ 28333] And some more DynaLoader things that should not be tested on VMS since they are not part of the implementation. [ 28840] darwin also seems to be lacking a dl_unload_file() in its DynaLoader, so some additional skips are needed. If I only had a NeXT to test with now... [ 29017] Upgrade to XSLoader-0.07. [ 29188] Subject: [patch at 21983] factor out $^O eq 'darwin' from XSLoader.pm From: "Konovalov, Vadim Vladimirovich (Vadim)** CTR **" Date: Thu, 2 Nov 2006 12:36:49 +0100 Message-ID: [ 29198] Wrap the $VERSION initialization of DynaLoader in a BEGIN block Subject: [perl #40651] (Obscure) Bug in Dynaloader From: gml4410@ggr.co.uk (via RT) Date: Thu, 02 Nov 2006 09:30:34 -0800 Message-ID: [ 29200] The assignment to $VERSION must be on its own line, for ExtUtils::MM_Unix::parse_version to be happy. [ 29253] The DynaLoader test isn't happy with OpenBSD's libc.so.39.3. [ 29580] Subject: [PATCH 5.8.8] OS/2-specific fixes, round II From: Ilya Zakharevich Date: Mon, 18 Dec 2006 00:45:24 -0800 Message-ID: <20061218084524.GA14866@powdermilk.math.berkeley.edu> [ 29707] Subject: DynaLoader test failure on cygwin From: "Yitzchak Scott-Thoennes" Date: Sun, 7 Jan 2007 11:56:21 -0800 (PST) Message-ID: <1675.63.226.247.31.1168199781.squirrel@63.226.247.31> [ 29857] Fix a typo in Dynaloader_pm.PL. Branch: maint-5.8/perl +> ext/DynaLoader/t/DynaLoader.t !> MANIFEST Makefile.SH ext/DynaLoader/DynaLoader_pm.PL !> ext/DynaLoader/Makefile.PL ext/DynaLoader/XSLoader_pm.PL !> ext/DynaLoader/dl_vms.xs ext/DynaLoader/dlutils.c !> ext/DynaLoader/t/XSLoader.t ext/util/make_ext !> lib/ExtUtils/Embed.pm t/lib/1_compile.t writemain.SH ____________________________________________________________________________ [ 30113] By: nicholas on 2007/02/04 12:08:57 Log: Revert the reversion of change 23524 and restore 23348: [ 23348] Subject: [PATCH-for-23341] dynaloader improvements and cleanup From: Vadim Konovalov Date: Sun, 3 Oct 2004 22:10:06 +0400 Message-ID: <138-1837306906.20041003221006@vkonovalov.ru> Branch: maint-5.8/perl ! ext/DynaLoader/DynaLoader_pm.PL ____________________________________________________________________________ [ 30112] By: nicholas on 2007/02/04 11:54:14 Log: Provide a fixed $DynaLoader::XS_VERSION of 1.05, to fix bug #32539 whilst also allowing $DynaLoader::VERSION to be incremented (and DynaLoader upgraded) Branch: maint-5.8/perl ! ext/DynaLoader/DynaLoader_pm.PL ext/DynaLoader/Makefile.PL ____________________________________________________________________________ [ 30110] By: nicholas on 2007/02/03 19:00:21 Log: Clarify that all the dire (sounding) "unsafe" warnings apply to 5005 threads. Branch: maint-5.8/perl ! gv.c op.c perl.c sv.c ____________________________________________________________________________ [ 30109] By: nicholas on 2007/02/03 18:52:20 Log: Integrate: [ 29502] Downgrading a fixed or floating substring of a pattern whilst matching a studied string seems to get to a "should not happen" [bug #41067] It seems that Perl_regexec_flags() assumes that if the pre-downgraded substring is FBM compiled, then the downgraded version will be too, hence changing the downgrade and upgrade routines to FBM compile seems to be a correct fix. [ 29503] Remove code duplication in S_to_utf8_substr() and S_to_byte_substr() by taking advantage of how anchored_* and float_* are stored in arrays to use a loop. [ 30092] Change 29502 wasn't perfect - you need to remove any extra trailing "\n" added by fbm_compile(), before recompiling with the same flags. In turn, to do that, it's best to store the flags even for short "PVBM"s. Branch: maint-5.8/perl ! regexec.c !> regcomp.h t/op/re_tests util.c ____________________________________________________________________________ [ 30107] By: nicholas on 2007/02/03 18:08:37 Log: Integrate: [ 26793] Cygwin issue (26792) and patchlevel (26790) Subject: Re: Cygwin status, maint and blead From: Yitzchak Scott-Thoennes Date: Fri, 6 Jan 2006 13:47:10 -0800 Message-ID: <20060106214710.GA592@efn.org> Subject: Re: Cygwin status, maint and blead From: Yitzchak Scott-Thoennes Date: Thu, 10 Nov 2005 00:46:33 -0800 Message-ID: <20051110084633.GA3364@efn.org> [ 26947] Make dump and -u croak on Win32 and Cygwin to avoid confusion such as was seen here: Subject: Weird Win32 error: "This application has requested the Runtime to terminate it in an unusual way." From: demerphq Date: Thu, 26 Jan 2006 15:40:42 +0100 Message-ID: <9b18b3110601260640x34b2442cnf5379d2f83b74187@mail.gmail.com> [ 27252] Subject: [PATCH] op/magic failure on cygwin after 1.5.19-4 From: Yitzchak Scott-Thoennes Date: Mon, 20 Feb 2006 11:09:33 -0800 Message-ID: <20060220190933.GA1316@efn.org> [ 27865] Subject: [PATCH] fix cygwin build From: Yitzchak Scott-Thoennes Date: Sun, 16 Apr 2006 18:05:20 -0700 Message-ID: <20060417010520.GA2888@efn.org> [ 28803] Fix to first problem causing Cygwin Perl to fail to build in a path with spaces in it. [ 29142] Changes to get perl to compile with g++ on Cygwin. Some additional changes will be needed to get it to link though. [ 29427] -Dprefix failed on Cygwin unless the directory ${installbin} already existed prior to running make. [ 29713] Subject: help cygwin find its dlls even on taint tests with no path From: "Yitzchak Scott-Thoennes" Date: Sun, 7 Jan 2007 16:56:28 -0800 (PST) Message-ID: <2437.63.226.247.31.1168217788.squirrel@63.226.247.31> Branch: maint-5.8/perl !> Configure cygwin/Makefile.SHs cygwin/cygwin.c malloc.c perl.c !> perl.h pod/perldiag.pod pod/perlport.pod sv.c t/op/magic.t !> t/op/taint.t ____________________________________________________________________________ [ 30106] By: nicholas on 2007/02/03 17:15:45 Log: Integrate: [ 28896] Subject: reentr reshuffle From: Jarkko Hietaniemi Date: Thu, 28 Sep 2006 12:40:04 +0300 Message-ID: <451B9874.7060000@iki.fi> [ 28897] Regen for #28896 [ 29639] With xlow and xhigh as U32, PAD_MAX shouldn't be IV_MAX, as on a 64 bit system this can't be stored. Branch: maint-5.8/perl ! reentr.h !> pad.c pod/perlembed.pod pod/perlxs.pod reentr.c reentr.pl ____________________________________________________________________________ [ 30102] By: nicholas on 2007/02/02 22:11:03 Log: Integrate: [ 29616] Fix RT #6006: Regexp replaces using large replacement variables fail some of the time (i.e. when the substitution contains something like ${10}). Patch derived from a patch by Hugo van der Sanden; added also a second test (Yves Orton already added a TODO test for this.) [ 29617] A better fix for RT #6006: revert change 29616, which was bogus, and add OP_SCOPE to the list of "dangerous" ops that might appear on the right side of s///. However, this fix does not work under taint mode, and op/subst.t has -T. [ 29618] Remove -T from op/subst.t, and add a new test file, substT.t, to run the s/// tests under taint mode; mark the test for RT #6006 as TODO, but only with -T. [ 29619] Fix #6006 for taint mode too. Branch: maint-5.8/perl +> t/op/substT.t ! op.c !> MANIFEST t/op/subst.t ____________________________________________________________________________ [ 30100] By: nicholas on 2007/02/02 21:52:26 Log: Integrate: [ 29308] Subject: [perl #36909] $^R undefined on matches involving backreferences From: yves orton via RT Date: Nov 17, 2006 4:07 PM Branch: maint-5.8/perl ! t/op/pat.t !> regcomp.c regexec.c t/op/subst.t ____________________________________________________________________________ [ 30099] By: nicholas on 2007/02/02 21:14:53 Log: Integrate: [ 29644] As freeing up PL_linestr is now done via the scope stack, there's no need to make any temporary varariable a mortal, and waste effort on the temp stack (not) clearing it up. [ 29647] Rationalise the logic in lex_start for creating a new temporary if the passed in SV is read only or doesn't end in ';'. Also, allow a NULL pointer to imply a zero length string, which saves creating a mortal in pp_require, only for it to be ignored in favour of a new SV. [ 29685] Subject: Re: [PATCH] Symbian port: add Series 90 support From: Jarkko Hietaniemi Date: Thu, 04 Jan 2007 03:35:54 -0500 Message-ID: <459CBC6A.4060806@iki.fi> Fixes a VC7/VC7.1 linker error following #29650/#29651 Branch: maint-5.8/perl !> embed.fnc perl.h pp_ctl.c proto.h toke.c util.c ____________________________________________________________________________ [ 30098] By: nicholas on 2007/02/02 20:54:46 Log: Integrate: [ 29193] eval $undef should emit one warning, not three. Also ensure that eval $undef clears $@ (it did, but only by luck) [ 29194] add test that eval undef clears $@ [ 29195] Change #29193 makes this addition unneeded Branch: maint-5.8/perl !> pp_ctl.c t/lib/common.pl t/op/eval.t toke.c ____________________________________________________________________________ [ 30097] By: nicholas on 2007/02/02 20:40:18 Log: Integrate: [ 26401] Missing file from last change [ 29182] Fix the failures in warnings tests when PERL_UNICODE is defined that show up in the smokes under UTF-8 locales. Based on : Subject: Re: UTF-8 Failures in smoke ($PERL_UNICODE) From: "H.Merijn Brand" Date: Wed, 1 Nov 2006 17:03:55 +0100 Message-ID: <20061101170355.7c57b515@pc09> [ 29184] More heuristics to make warnings.t pass under different combinations of PERL_UNICODE / locale Branch: maint-5.8/perl +> t/lib/common.pl !> MANIFEST lib/warnings.t ____________________________________________________________________________ [ 30096] By: nicholas on 2007/02/02 18:03:45 Log: Integrate: [ 28559] Cleanup and conversion to t/test.pl [ 29643] PL_linestr needs to survive until the end of scope, not just the next FREETMPS. Fixes the underlying cause of the thread cloning SEGV reported in http://www.nntp.perl.org/group/perl.perl5.porters/63123 [ 29668] PVMG can be isUV too. Branch: maint-5.8/perl +> t/op/threads_create.pl ! dump.c !> MANIFEST ext/Devel/Peek/t/Peek.t sv.c t/op/threads.t toke.c ____________________________________________________________________________ [ 30095] By: nicholas on 2007/02/02 17:46:17 Log: Integrate: [ 29628] Change PAD_MAX to something a bit less FORTRAN-ish than 999999999. [ 29629] Abstract the pad code's overloaded use of SvNVX and SvIVX into 4 macros COP_SEQ_RANGE_LOW, COP_SEQ_RANGE_HIGH, PARENT_PAD_INDEX and PARENT_FAKELEX_FLAGS [ 29631] Move PAD_FAKELEX_ANON and PAD_FAKELEX_MULTI to pad.h, export them via B.pm, so that B::Concise doesn't need to hard code magic numbers. [ 29663] Probably better to make PAD_FAKELEX_ANON and PAD_FAKELEX_MULTI special cases, instead of scanning pad.h to find them, and picking up 3 other superfluous definitions as a side effect. Branch: maint-5.8/perl ! pad.c !> ext/B/B.pm ext/B/B.xs ext/B/B/Concise.pm ext/B/B/Deparse.pm !> ext/B/defsubs_h.PL ext/B/t/concise-xs.t pad.h ____________________________________________________________________________ [ 30094] By: nicholas on 2007/02/02 16:57:24 Log: Integrate: [ 29506] Subject: [PATCH] Re: [perl #41065] Out of memory!, while extending scalar From: Marcus Holland-Moritz Date: Sun, 10 Dec 2006 22:32:32 +0100 Message-ID: <20061210223232.0f3a5318@r2d2> [ 29574] Subject: [PATCH] perl.c: leak avoidance From: Jarkko Hietaniemi Date: Sat, 16 Dec 2006 11:48:13 -0500 Message-ID: <4584234D.7070209@iki.fi> [ 29590] Subject: [PATCH] Don't destroy the Unicode system environment on Perl startup From: Jan Dubois Date: Mon, 18 Dec 2006 16:34:33 -0800 Message-ID: [ 29614] syswrite() wasn't reporting a warning when writing to a filehandle opened only for input. Branch: maint-5.8/perl !> doop.c perl.c pp_sys.c t/lib/warnings/pp_sys ____________________________________________________________________________ [ 30077] By: nicholas on 2007/01/29 23:50:30 Log: Integrate: [ 29492] In Perl_fbm_instr(), no need to calculate the address of the table if we're going to return anyway. [ 29493] Exterminate! 2 pieces of superfluous code related to UV setting. [ 29494] PVBMs mustn't enter anything that turns SVIVisUV_on(). [ 29518] Change SvTAIL() to check that both SVpbm_TAIL|SVpbm_VALID are true. SVpbm_VALID is the same bit value is SVf_IVisUV, which means that PVBMs can't actually ever be IOK. Therefore move BmUSEFUL() into the IV union, and save one I32 per PVBM. [ 29534] Shrink Perl_sv_peek() and Perl_do_sv_dump() by using type to name lookup tables. Branch: maint-5.8/perl ! dump.c !> sv.c util.c ____________________________________________________________________________ [ 30075] By: nicholas on 2007/01/29 23:16:13 Log: Integrate: [ 29429] Verify that the debugger has an array where to store lines before doing so. This fixes an assertion failure when parsing a script that begins with '#!perl -d'. Also, code factorization in toke.c. [ 29433] Remove the pp_entersub specific logic from S_get_db_sub. [ 29434] Make get_db_sub non-static, and call it from pp_goto, which allows the removal of duplicate code. (The conversion of GvSV(PL_DBsub) to GvSVn(PL_DBsub) implicit in this change should fix a failure with Devel::SmallProf.) [ 29463] Move Perl_get_db_sub() from pp_hot.c to util.c [ 29749] In toke.c, merge S_update_debugger_info_pv() and S_update_debugger_info_sv() into S_update_debugger_info(). Branch: maint-5.8/perl ! util.c !> embed.fnc embed.h op.c pp_ctl.c pp_hot.c proto.h toke.c ____________________________________________________________________________ [ 30074] By: nicholas on 2007/01/29 22:57:54 Log: Integrate: [ 29327] Make undef $SIG{FOO} ignore the signal FOO [ 29328] Actually, undef $SIG{FOO} should restore the default signal handler. Else, once we exit a block where the signal handler was localized, it becomes ignored. Branch: maint-5.8/perl !> mg.c ____________________________________________________________________________ [ 30073] By: nicholas on 2007/01/29 22:44:03 Log: Integrate: [ 29311] Subject: [perl #18209] bug in regex s/\s$word\s/ /gi; and $& From: "yves orton via RT" Date: Fri, 17 Nov 2006 10:17:48 -0800 Message-ID: [ 29312] Subject: [perl #18209] bug in regex s/\s$word\s/ /gi; and $& From: "yves orton via RT" Date: Fri, 17 Nov 2006 10:17:48 -0800 Message-ID: Branch: maint-5.8/perl ! t/op/pat.t !> mg.c ____________________________________________________________________________ [ 30071] By: nicholas on 2007/01/29 22:30:00 Log: Integrate: [ 29361] Fix for bug #38631: tied variables don't work with .= <> [ 29416] Subject: [perl #41008] Setting $0 invalidates environment shown by ps From: alexander_bluhm@genua.de (via RT) Date: Tue, 28 Nov 2006 05:09:31 -0800 Message-ID: [ 29446] Subject: Re: [perl #38868] Changing $0 on darwin leads to excessive padding in 'ps' From: Chris Dolan Date: Sat, 2 Dec 2006 09:37:08 -0600 Message-Id: <005671A0-5C2B-464C-85B9-A91E52278194@chrisdolan.net> [ 29488] add missing OP_REFCNT_LOCK/UNLOCKs and document it [ 29530] The overly-picky AIX xlc compiler doesn't like the style from patch #29446 [ 29531] Subject: [PATCH] mg.c: #ifdef only the different bits From: jhi@cc.hut.fi (Jarkko Hietaniemi) Date: Tue, 12 Dec 2006 15:18:53 +0200 (EET) Message-Id: <20061212131853.C2CBA43A67@anubis.hut.fi> Branch: maint-5.8/perl !> mg.c op.h pp_hot.c sv.c t/op/readline.t ____________________________________________________________________________ [ 30070] By: nicholas on 2007/01/29 22:12:03 Log: Integrate: [ 29239] [perl #40718] perl parser bug leading to memory corruption quoted-string parser naughtily maintained a pointer into an SV which could get realloc()ed. [ 29263] Fix a bug in the debugger tracing variables when one was eval'ing a string including a "#line" directive containing the actual name of the file already. (following-up change #25409) [ 29273] Fix bug #39037: rcatline doesn't stringify references [ 29310] Subject: [perl #22395] regexp /(.*)[bc]/ 10000 times slower in 5.8.0 vs 5.6.1 From: "yves orton via RT" Date: Fri, 17 Nov 2006 09:48:14 -0800 Message-ID: [ 29310] Subject: [perl #22395] regexp /(.*)[bc]/ 10000 times slower in 5.8.0 vs 5.6.1 From: "yves orton via RT" Date: Fri, 17 Nov 2006 09:48:14 -0800 Message-ID: [ 29310] Subject: [perl #22395] regexp /(.*)[bc]/ 10000 times slower in 5.8.0 vs 5.6.1 From: "yves orton via RT" Date: Fri, 17 Nov 2006 09:48:14 -0800 Message-ID: [ 29330] Unify the two warnings "Can't declare %s in %s", in line with what's documented in perldiag. [ 29336] Invalidate the method lookup cache when assigning to a glob named "isa". (That happens when importing "isa" from UNIVERSAL, for example.) Fixes bug #24824. [ 29350] Fix infinite loop in Perl_my_strftime() for failing strftime() Branch: maint-5.8/perl ! t/op/pat.t !> op.c pp_hot.c regcomp.c t/op/attrs.t t/op/readline.t !> t/op/universal.t toke.c util.c ____________________________________________________________________________ [ 30069] By: nicholas on 2007/01/29 21:05:26 Log: Integrate: [ 29201] Subject: [PATCH] replace S_sv_pos_b2u_forwards with utf8_length (was: sv_pos_b2u dislikes the extended UTF-8) From: SADAHIRO Tomoyuki Date: Sun, 21 May 2006 14:00:43 +0900 Message-Id: <20060521140011.358D.BQW10602@nifty.com> [ 29229] Subject: [PATCH] Perl_die() / Perl_croak() From: "Philip M. Gollucci" Date: Tue, 07 Nov 2006 23:27:40 -0800 Message-ID: <455186EC.8010906@p6m7g8.com> (backported to embed.fnc) [ 29259] Subject: [PATCH] was RE: Perl_die() / Perl_croak() From: "Robin Barker" Date: Mon, 13 Nov 2006 10:25:08 -0000 Message-ID: <2C2E01334A940D4792B3E115F95B7226149377@exchsvr1.npl.ad.local> [ 29313] Tweaks needed to get Perl compiling with g++ on OpenBSD. [ 29363] Remove superfluous SvROK() [ 29386] Change (well, correct) S_add_data to take and return a U32. [ 29387] Simplify S_add_data(), given that realloc will NULL acts as malloc(). [ 29476] save_I8 is not a mathom. [ 29482] Subject: [PATCH] Perl_save_destructor is not a mathom From: "Jerry D. Hedden" Date: Wed, 6 Dec 2006 11:12:13 -0800 (PST) Message-ID: <20061206191213.35036.qmail@web30206.mail.mud.yahoo.com> [ 29486] No need to upgrade if all we're going to do is return. Branch: maint-5.8/perl ! regcomp.c scope.c sv.c !> embed.fnc embed.h embed.pl mathoms.c perl.h perlio.c proto.h !> t/op/index.t util.c ____________________________________________________________________________ [ 30068] By: nicholas on 2007/01/29 20:23:46 Log: Integrate: [ 29197] When code is loaded through an @INC-hook, and when this hook has set a filename entry in %INC, make sure __FILE__ is set for this code accordingly to the contents of that %INC entry. [ 29235] Change the documentation of the return values of @INC-hooks to match what is tested to work. (It's a bit confusing that optional values might appear at the middle, too.) [ 29236] Subject: [PATCH t/op/inccode.t] fails under minitest From: "Robin Barker" Date: Wed, 8 Nov 2006 15:18:23 -0000 Message-ID: <2C2E01334A940D4792B3E115F95B7226120AA5@exchsvr1.npl.ad.local> [ 29584] Subject: [PATCH blead] Re: [perl #41071] require stringifies code references in tied @INC From: Rick Delaney Date: Fri, 15 Dec 2006 23:28:25 -0500 Message-ID: <20061216042825.GB23501@localhost.localdomain> Branch: maint-5.8/perl +> t/op/inccode-tie.t !> MANIFEST pod/perlfunc.pod pp_ctl.c t/op/inccode.t ____________________________________________________________________________ [ 30066] By: nicholas on 2007/01/29 19:07:36 Log: Integrate: [ 29033] Subject: [perl #40557] regexec.c saves context stack position improperly From: Dave Bailey (via RT) Date: Tue, 17 Oct 2006 09:37:28 -0700 Message-ID: [ 29059] zero length $1 took too long to detect utf8-ness [ 29063] only do \C hackery if \C is seen From: demerphq Date: Fri, 20 Oct 2006 12:38:13 +0200 Subject: Re: utf8 regexp performance problem Message-ID: <9b18b3110610200338k5df1d3afpf829fbc0f3663a75@mail.gmail.com> Branch: maint-5.8/perl !> mg.c regexec.c scope.c scope.h ____________________________________________________________________________ [ 30065] By: nicholas on 2007/01/29 18:52:30 Log: Integrate: [ 29040] panic if we attempt to sv_setsv to or from an already freed SV. [ 29097] No warning was produced if perl failed to load PerlIO::scalar (contrary to other PerlIO layers.) [ 29108] Fix small leak with -F at interpreter destruction [ 29116] Subject: Re: [perl #40583] sprintf "%#04X" also uppercases the 0x-prefix From: SADAHIRO Tomoyuki Date: Fri, 27 Oct 2006 00:19:34 +0900 Message-Id: <20061027001924.B158.BQW10602@nifty.com> [ 29143] Change the non-mkstemp, non-win32 code path of PerlIO_tmpfile to use PerlIO_fdopen so that when the tmpfile is closed we do not decrement a ref count that doesn't exist or is zero. Branch: maint-5.8/perl !> perl.c perlio.c pod/perlfunc.pod sv.c t/op/sprintf.t ____________________________________________________________________________ [ 30063] By: nicholas on 2007/01/29 18:20:31 Log: And now actually fixed. (All tests pass) Branch: maint-5.8/perl ! sv.c ____________________________________________________________________________ [ 30062] By: nicholas on 2007/01/29 18:02:23 Log: Fix up damage from change 30061 - at least make sv.c compile. [Problem caused by not running make test in the correct tree] Branch: maint-5.8/perl ! sv.c ____________________________________________________________________________ [ 30061] By: nicholas on 2007/01/29 17:39:20 Log: Integrate: [ 29008] Don't bother checking for bad characters in prototypes if we're not going to warn about them. [ 29028] Less brace nesting. [ 29042] Using explicit printf-like APIs with a format of "%s" isn't the world's most efficient idea. [ 29068] Quiet a warning in pp_sys.c on AIX. [ 29157] VOL should be used instead of volatile in the core. [ 29190] This volatile modifier is not on the referent, but on the pointer (see change 28606) Branch: maint-5.8/perl !> XSUB.h dump.c op.c perl.c pp_sys.c toke.c ____________________________________________________________________________ [ 30060] By: nicholas on 2007/01/29 17:14:16 Log: Integrate: [ 28867] Subject: [PATCH] C++: add -Wno-used-parameter, and drop ODBM_File From: Jarkko Hietaniemi Date: Tue, 19 Sep 2006 09:19:22 +0300 Message-ID: <450F8BEA.1010802@iki.fi> [ 28914] Subject: Re: [PATCH] cflags.SH: rethink of the gcc -std=c89 and -pedantic From: Jarkko Hietaniemi Date: Mon, 02 Oct 2006 13:04:14 +0300 Message-ID: <4520E41E.8040300@iki.fi> [ 28955] Subject: [PATCH] cflags.SH: strip -std=c89 for g++ From: Jarkko Hietaniemi Date: Sat, 07 Oct 2006 13:37:04 +0300 Message-ID: <45278350.8020707@iki.fi> [ 29398] Subject: [PATCH] Re: [PATCH] perlhack: some portability updates From: Dominic Dunlop Date: Sun, 26 Nov 2006 12:01:16 +0100 Message-Id: <253514EB-BA57-4A43-93FA-75D6F3CF27BC@mac.com> [ 29400] Typo fix (by Dominic Dunlop) [ 30016] If we have specified -Dgccansipedantic on the Configure command-line, assume we really want it [ 30019] cflags.SH was broken: it was compiling a C file for testing available command-line options, but this C file was including perl.h, which in turn includes config.h, which might not be present at that time. So force the generation of config.h. Branch: maint-5.8/perl !> Makefile.SH cflags.SH perl.h ____________________________________________________________________________ [ 30058] By: nicholas on 2007/01/29 16:46:38 Log: Integrate: [ 28917] Subject: [perl #40427] Segfault in pack From: dgay@acm.org (via RT) Date: Thu, 28 Sep 2006 17:30:37 -0700 Message-ID: [ 28975] fix bad free in do_exec3() [ 28985] Subject: [perl #40473] sprintf width+precision fails on wide chars From: Anatoly Vorobey (via RT) Date: Sun, 08 Oct 2006 17:58:16 -0700 Message-ID: [ 28993] Missing % in format string [ 29016] Avoid undefined behaviour for -DPERL_MEM_LOG by not using a direct dereference of member of the structure being reallocated as part of the size calculation. (There may be other similar bugs). [ 29025] Subject: [PATCH] Re: sprintf 64 test From: SADAHIRO Tomoyuki Date: Sun, 15 Oct 2006 16:51:34 +0900 Message-Id: <20061015165052.77AD.BQW10602@nifty.com> 1. nullify 0 flag in integer conversions when precision is given 2. ignore space after a plus sign as a sign for a nonnegative number 3. make a negative precision through * working as if the precision is omitted Branch: maint-5.8/perl !> doio.c perl.h pod/perlfunc.pod pp_pack.c regcomp.c sv.c !> t/op/sprintf.t t/op/sprintf2.t ____________________________________________________________________________ [ 30057] By: nicholas on 2007/01/29 15:55:07 Log: Integrate: [ 28877] Subject: [PATCH] deal with some gcc warnings From: Jarkko Hietaniemi Date: Thu, 21 Sep 2006 09:08:47 +0300 Message-ID: <45122C6F.9080904@iki.fi> [ 28910] In pp_binmode, call mode_from_discipline() once and remember the result. [ 28911] In PerlIO_debug(), if tainting or set*id, set PL_perlio_debug_fd to -1 first time through to reduce the checks on subsequent calls. [ 28947] Don't bother generating the "Operation \"%s\": no method found..." message text if we're not going to use it. [ 28950] Generate the overload enum and names array programatically, which allows experimentation with the order. The new order shaves nearly 900 bytes from gv.o, because the compiler can make smaller branch tables for switch statements. [ 28951] Fix typo spotted by Rafael. Close the file handle explicity and check for errors. Add overload.pl to regen.pl [ 28952] Make the executable slightly smaller by using PL_hexdigit in Perl_sv_vcatpvfn. [ 28961] Make reference stringification (blessed and unblessed) about as fast as is possible, because I'm told it's used quite frequently. [ 29015] Use Perl_croak_nocontext() rather than Perl_croak() for the snprintf() and vsnprintf() macros. Branch: maint-5.8/perl + overload.h overload.pl !> MANIFEST doio.c gv.c perl.h perlio.c pp_sys.c reentr.c !> reentr.pl regen.pl sv.c util.c ____________________________________________________________________________ [ 30056] By: nicholas on 2007/01/29 11:15:51 Log: Integrate: [ 27795] (Hopefully) fix coredumps in ByteLoader, which were a side effect of fixing a bug in B, and non-robust autogenerated code. When autogenerating initialiser statements for an array, it's a bad plan to have the size of that array hard coded. [ 27826] Coverity was getting upset about an assignment from a function returning int to a U8 variable. Curiously it was only getting upset about the first of many. However, we are doing NO ERROR CHECKING on whether we read in the requested number of bytes of bytecode. All except BGET_op_tr_array will now croak on a short read. [ 27874] Oops. Change 27826 had an error in the definition of BGET_PADOFFSET(). [ 28638] Silence the VC7 compiler warning in the following smoke: http://www.nntp.perl.org/group/perl.daily-build.reports/39824 [ 28667] Try to reinstate the fix of change #28638 by patching bytecode.pl and not the file it generates [ 28668] Provide a BGET_svtype, required by #28667 [ 28739] Subject: [PATCH] bytecode.t failures in Tru64: svtype (enum) is int, not long From: Jarkko Hietaniemi Date: Sat, 19 Aug 2006 14:34:53 +0300 (EEST) Message-Id: <200608191134.k7JBYroK518962@kosh.hut.fi> plus fix 28739 to use U32 rather than I32 to avoid an error on this ILP 32 bit system. (Checked also with -Duse64bitint, but we await the verdict on an LP64 system) Branch: maint-5.8/perl ! ext/B/B/Assembler.pm !> bytecode.pl ext/B/B/Asmdata.pm ext/ByteLoader/bytecode.h !> ext/ByteLoader/byterun.c ____________________________________________________________________________ [ 30055] By: nicholas on 2007/01/28 23:54:52 Log: Integrate: [ 28297] Subject: [PATCH] Stop harness from printing summary table header for each row in table From: demerphq Date: Tue, 23 May 2006 22:27:40 +0200 Message-ID: <9b18b3110605231327t316104caoce301668178ee442@mail.gmail.com> [ 28304] Upgrade to Test-Harness-2.60 [ 28384] Upgrade to Test::Harness 2.62 [ 28386] Adapt the new Test::Harness test to the core [ 28953] Upgrade to Test-Harness-2.64 Branch: maint-5.8/perl +> lib/Test/Harness/Results.pm lib/Test/Harness/t/failure.t !> MANIFEST lib/Test/Harness.pm lib/Test/Harness/Changes !> lib/Test/Harness/Straps.pm lib/Test/Harness/Util.pm !> lib/Test/Harness/bin/prove lib/Test/Harness/t/00compile.t !> lib/Test/Harness/t/callback.t !> lib/Test/Harness/t/prove-switches.t !> lib/Test/Harness/t/strap-analyze.t !> lib/Test/Harness/t/test-harness.t ____________________________________________________________________________ [ 30054] By: nicholas on 2007/01/28 23:29:04 Log: Change 28489 giveth, and change 28488 taketh away. (Although for most people time flows forwards, so they won't see it this way). But the problem is that 28488 taketh away the {}, and whilst 28489 puteth them back, 28489 was about state variables, so I'm not merging it. Hence this need for a manual fixup. (And something else I'd like to fixup is gcc, for not having a C89-death-or-glory flag) Branch: maint-5.8/perl ! op.c ____________________________________________________________________________ [ 30053] By: nicholas on 2007/01/28 23:11:59 Log: Fix bug introduced by a clean merge in change 30051. Either my test environment is screwy, or my ability to read (or at least notice things) is impaired. Although I suspect upgrading to Test::Harness 2.62 will help, as that restores the die-on-errors, which t/harness lacks currently. Branch: maint-5.8/perl ! ext/B/B.xs ____________________________________________________________________________ [ 30051] By: nicholas on 2007/01/28 21:56:48 Log: Integrate: [ 28626] Clear up most of the VC6 compiler warnings in the following smoke: http://www.nntp.perl.org/group/perl.daily-build.reports/39733 Branch: maint-5.8/perl !> ext/B/B.xs ext/XS/APItest/APItest.xs hv.c op.c pp.c pp_pack.c !> utf8.c ____________________________________________________________________________ [ 30049] By: nicholas on 2007/01/28 08:28:43 Log: Update Changes Branch: maint-5.8/perl ! Changes patchlevel.h ____________________________________________________________________________ [ 30048] By: nicholas on 2007/01/28 00:08:17 Log: Integrate: [ 27951] Subject: [perl #38709] Opening '|-' triggers unjustified taint check From: mhasch@cpan.org (via RT) Date: Fri, 10 Mar 2006 20:10:49 -0800 Message-ID: [ 28488] Fix a bug on setting OPpASSIGN_COMMON on a AASSIGN op when the left side is made out a list declared with our(). In this case OPpLVAL_INTRO isn't set on the left op, so we just remove that check. Add new tests. [ 28879] Subject: RE: Combining UTF-16 output with :crlf is awkward From: "Jan Dubois" Date: Thu, 6 Apr 2006 18:37:21 -0700 Message-ID: <0bbe01c659e3$d0c334a0$2217a8c0@candy> [ 29297] Change #27951 added tests to t/op/taint.t involving taint, opening "|-". Unfortunately, this usually reacts badly on OpenBSD with threaded Perls, causing zombie processes nearly every time this is run. For now, skipping the test seem reasonable while working on a solution to the problem. [ 29302] Change 29297 omitted a semicolon. Branch: maint-5.8/perl !> doio.c op.c perlio.c t/op/array.t t/op/taint.t ____________________________________________________________________________ [ 30047] By: nicholas on 2007/01/27 23:49:02 Log: Integrate: [ 28359] Upgrade to Encode 2.18 Branch: maint-5.8/perl !> ext/Encode/Byte/Byte.pm ext/Encode/CN/CN.pm ext/Encode/Changes !> ext/Encode/EBCDIC/EBCDIC.pm ext/Encode/Encode.pm !> ext/Encode/Encode.xs ext/Encode/JP/JP.pm ext/Encode/KR/KR.pm !> ext/Encode/Symbol/Symbol.pm ext/Encode/TW/TW.pm !> ext/Encode/bin/enc2xs ext/Encode/encoding.pm !> ext/Encode/lib/Encode/Alias.pm !> ext/Encode/lib/Encode/CJKConstants.pm !> ext/Encode/lib/Encode/CN/HZ.pm ext/Encode/lib/Encode/Config.pm !> ext/Encode/lib/Encode/Encoding.pm !> ext/Encode/lib/Encode/Guess.pm ext/Encode/lib/Encode/JP/H2Z.pm !> ext/Encode/lib/Encode/JP/JIS7.pm !> ext/Encode/lib/Encode/KR/2022_KR.pm !> ext/Encode/lib/Encode/MIME/Header.pm !> ext/Encode/lib/Encode/MIME/Header/ISO_2022_JP.pm !> ext/Encode/lib/Encode/Unicode/UTF7.pm ____________________________________________________________________________ [ 30046] By: nicholas on 2007/01/27 23:25:32 Log: Integrate: [ 26958] Subject: [PATCH] quiet gcc Info message in cygwin From: Yitzchak Scott-Thoennes Date: Wed, 25 Jan 2006 19:16:09 -0800 Message-ID: <20060126031608.GA2588@efn.org> [ 27212] Don't add -fPIC if already present. Subject: [PATCH] 5.8.8: Debian patches From: "Brendan O'Dea" Date: Sat, 11 Feb 2006 00:37:15 +1100 Message-ID: <20060210133715.GA6826@londo.c47.org> [ 27250] Subject: [PATCH] Make SDBM_File work with -Duse64bitall on Darwin (Mac OS X) From: Dominic Dunlop Date: Feb 20, 2006 1:46 PM Message-Id: [ 27333] Subject: [PATCH] Suppress System V IPC for 64-bit Darwin builds From: Dominic Dunlop Date: Fri, 24 Feb 2006 17:22:54 +0100 Message-Id: [ 27995] Anton Berezin says that on FreeBSD we're wrong to be using -lc_r, and should just be using -pthread on the linker line. This stops the new taint tests looping infinitely. It also seems to cure the TODO test in posix.t - masked SIGINT is now received. [ 28116] Remvove code that gets undone in the next line. Keep the documentation part. [ 28155] Add -fno-delete-null-pointer-checks to ccflags on Itaniums when compiling with gcc's 3.2, 3.3, or 3.4. This fixes RT #37156. [ 28309] Sun C on Linux doesn't complain about the attributes sprinkled through the code in all cases, even though attributes are not supported. In some cases, however, it does. [ 28446] Solaris hints: gcc flags confuse Configure Subject: [PATCH] (Was Re: [perl #39636] Configure fails) From: Andy Dougherty Date: Wed, 28 Jun 2006 13:37:54 -0400 (EDT) Message-ID: [ 28458] Solaris' sh doesn't like empty if's [ 28539] Subject: [PATCH] z/OS: hints/os390.sh and Makefile.SH From: Jarkko Hietaniemi Date: Mon, 10 Jul 2006 22:55:33 +0300 Message-ID: <44B2B0B5.3000008@iki.fi> [ 28751] Subject: [perl #40161] ccversion fix for hpux From: smoketst@hp46t243.cup.hp.com (via RT) Date: Wed, 16 Aug 2006 23:10:52 -0700 Message-ID: [ 28783] Subject: [perl #40239] New hints file for MidnightBSD From: Lucas Holt (via RT) Date: Fri, 25 Aug 2006 20:16:40 -0700 Message-ID: [ 28842] Subject: [PATCH] (1) g++ 4.x (or Mac OS X) (2) Solaris g++ or CC From: Jarkko Hietaniemi Date: Wed, 13 Sep 2006 20:19:04 +0300 Message-ID: <45083D88.7050207@iki.fi> Plus a tweak to the name of CC. [ 28979] Subject: [PATCH] Linux and Solaris hints: C++ vs dlerror From: Jarkko Hietaniemi Date: Tue, 10 Oct 2006 14:00:40 +0300 Message-ID: <452B7D58.1090009@iki.fi> [ 29334] Remove description of WHOA THERE messages for recommended hint values, removed by change #29288. [ 29335] Remove warnings against WHOA THERE warnings from the hints files. [ 29596] Subject: [PATCH] Re: 'localtime' not thread-safe under HP-UX B.11.11/64 From: "Jerry D. Hedden" Date: Mon, 18 Dec 2006 10:18:28 -0800 (PST) Message-ID: <31291.28361.qm@web30205.mail.mud.yahoo.com> [ 29777] gcc -dumpversion is at least supported back to 2.7.x and thus seems to be much more defensive against upcoming version reporting format changes. [ 29845] Reentrant functions appear to still be broken in NetBSD 3.1. Branch: maint-5.8/perl +> hints/midnightbsd.sh !> (integrate 27 files) ____________________________________________________________________________ [ 30045] By: nicholas on 2007/01/27 23:04:25 Log: Integrate: [ 28609] Subject: [PATCH] Tru64: if cc given both -O and -g, -g must be -g3 From: Jarkko Hietaniemi Date: Sun, 23 Jul 2006 10:18:11 +0300 (EEST) Message-Id: <200607230718.k6N7IB4u009170@kosh.hut.fi> [ 28633] Subject: [PATCH] (back) towards g++ compilation (in Tru64, and elsewhere) From: Jarkko Hietaniemi Date: Mon, 31 Jul 2006 23:44:55 +0300 Message-ID: <44CE6BC7.8060805@iki.fi> Just the hints/dec_osf.sh changes from this set of patches. [ 28654] Subject: [PATCH] more g++ in Tru64 From: Jarkko Hietaniemi Date: Thu, 03 Aug 2006 09:04:45 +0300 Message-ID: <44D191FD.9030902@iki.fi> [ 28655] Subject: [PATCH] Tru64 oops From: Jarkko Hietaniemi Date: Thu, 3 Aug 2006 22:59:37 +0300 (EEST) Message-Id: <200608031959.k73JxbsC395856@kosh.hut.fi> [ 28689] Subject: [PATCH] g++: gethostbyaddr, and more Tru64 defines From: Jarkko Hietaniemi Date: Thu, 10 Aug 2006 09:00:44 +0300 Message-ID: <44DACB8C.1080708@iki.fi> [ 28741] Subject: [PATCH] g++/Tru64: got perl linking once got the linkage right From: Jarkko Hietaniemi Date: Fri, 18 Aug 2006 13:03:17 +0300 (EEST) Message-Id: <200608181003.k7IA3Hso276655@kosh.hut.fi> Branch: maint-5.8/perl !> hints/dec_osf.sh perl.h perlio.c pp_sys.c ____________________________________________________________________________ [ 30042] By: nicholas on 2007/01/27 22:45:18 Log: Integrate: [ 28731] gcc-4.1.1 on Fedora optimized away NULL chacks in Perl_do_close(), which caused core dumps. [ 29753] do_gv_dump, do_gvgv_dump and do_hv_dump call all actually take a NULL parameter. Patch for do_hv_dump supplied by Brandon L Black, other two added after code inspection. [ 29758] Change 29753 messed up do_hv_dump() [my fault] Branch: maint-5.8/perl !> embed.fnc proto.h ____________________________________________________________________________ [ 30040] By: nicholas on 2007/01/27 18:56:32 Log: Integrate: [ 28759] Silence a bunch of Borland compiler warnings in SDBM_File (See: http://www.nntp.perl.org/group/perl.daily-build.reports/40283) [ 28786] Subject: [PATCH perl-current]; was RE: [PATCH perlio.c pp.c regcomp.c] silence format and comparison warnings From: "Robin Barker" Date: Fri, 1 Sep 2006 16:10:25 +0100 Message-ID: <2C2E01334A940D4792B3E115F95B7226120A21@exchsvr1.npl.ad.local> [ 28793] cxtype is a U16, not a U32 (Stops VC++ complaining) [ 28813] Subject: NULL and sv_setpv vs. newSVpv From: "Yitzchak Scott-Thoennes" Date: Sun, 10 Sep 2006 20:31:46 -0700 (PDT) Message-ID: <8750.24.19.24.170.1157945506.squirrel@24.19.24.170> [ 28863] Fix one genuine unitialised code path spotted by gcc, and a signed/ unsigned warning. [ 28865] Subject: [PATCH] bleed - silence warning in sv.c From: "Philip M. Gollucci" Date: Mon, 18 Sep 2006 01:30:05 -0700 Message-ID: <450E590D.1060205@p6m7g8.com> [ 28886] Subject: Re: Change 28877: [PATCH] deal with some gcc warnings From: Jarkko Hietaniemi Message-ID: <45175CA4.3020900@iki.fi> Date: Mon, 25 Sep 2006 07:35:48 +0300 remove casts that were causing failure on FreeBSD -Duse64bitint, and some other cleanups. [ 28899] Silence two warnings from gcc when being -pedantic Branch: maint-5.8/perl !> XSUB.h doio.c dump.c perl.h perlio.c pp.c pp_ctl.c pp_pack.c !> reentr.c reentr.pl sv.c utf8.c ____________________________________________________________________________ [ 30039] By: nicholas on 2007/01/27 18:27:17 Log: Integrate: [ 28753] Subject: [PATCH] Don't promulgate Perl attributes From: "Jerry D. Hedden" Date: Wed, 23 Aug 2006 10:07:05 -0700 Message-ID: <20060823100705.fb30e530d17747c2b054d625b8945d88.87c0ee9326.wbe@email.secureserver.net> [ 28756] Subject: [PATCH] RE: [perl #40227] 'reserved' warning not working From: "Jerry D. Hedden" Date: Thu, 24 Aug 2006 09:04:12 -0700 Message-ID: <20060824090412.fb30e530d17747c2b054d625b8945d88.e8ae12ccab.wbe@email.secureserver.net> Branch: maint-5.8/perl !> lib/attributes.pm xsutils.c ____________________________________________________________________________ [ 30038] By: nicholas on 2007/01/27 18:13:24 Log: Integrate: [ 28670] Fixups for test failures caused by #28667 and #28668 [ 28729] Additional changes to get C++ a little closer to a clean compile of Perl. [ 28730] Cast needed to compile bsd_glob.c with C++ [ 28744] Subject: [PATCH] g++: fix Digest::MD5, Math::BigInt::FastCalc, ODBM_File, XS::APItest (and Time::HiRes 1.88 in CPAN) From: Jarkko Hietaniemi Date: Tue, 22 Aug 2006 10:45:41 +0300 (EEST) Message-Id: <200608220745.k7M7jfjP050939@kosh.hut.fi> ...plus various version bumps and Digest::MD5 test file fixes. [ 28754] Subject: [PATCH ext/ODBM_File/ODBM_File.xs] change 28744 broke ODBM_File From: "Robin Barker" Date: Thu, 24 Aug 2006 13:22:47 +0100 Message-ID: <2C2E01334A940D4792B3E115F95B7226120A1A@exchsvr1.npl.ad.local> [ 28757] Following Jarkko's advice, it seems that ODBM_File cannot be compiled as C++ as-is. Branch: maint-5.8/perl !> bytecode.pl ext/B/B/Asmdata.pm ext/B/B/Assembler.pm !> ext/B/t/assembler.t ext/ByteLoader/bytecode.h !> ext/ByteLoader/byterun.c ext/DB_File/DB_File.pm !> ext/DB_File/DB_File.xs ext/Digest/MD5/MD5.pm !> ext/Digest/MD5/MD5.xs ext/Digest/MD5/t/files.t !> ext/File/Glob/bsd_glob.c ext/ODBM_File/ODBM_File.pm !> ext/ODBM_File/ODBM_File.xs ext/POSIX/POSIX.pm !> ext/POSIX/POSIX.xs ext/XS/APItest/APItest.pm !> ext/XS/APItest/APItest.xs ____________________________________________________________________________ [ 30036] By: nicholas on 2007/01/27 17:35:47 Log: Integrate: [ 26593] Subject: [patch@26561] utf8.c const issue. From: "John E. Malmberg" Date: Mon, 02 Jan 2006 14:16:02 -0500 Message-ID: <43B97BF2.2050201@qsl.net> [ 28733] syscall() declaration in perl.h needs to have an EXTERN_C prefixing it for some C++ compiles. Might as well fix usleep()'s declaration too. [ 28734] Subject: [PATCH] Re: A large-ish C++ TODO (Re: [PATCH] g++ stage 1 reached) From: Jarkko Hietaniemi Date: Thu, 17 Aug 2006 19:16:09 +0300 Message-ID: <44E49649.9090307@iki.fi> [ 28737] Subject: [PATCH] g++/Solaris: force d_dlopen=define From: Jarkko Hietaniemi Date: Fri, 18 Aug 2006 15:48:34 +0300 (EEST) Message-Id: <200608181248.k7ICmY2W028375@vipunen.hut.fi> [ 28742] Subject: [PATCH] g++/x2p (better version) From: Jarkko Hietaniemi Date: Mon, 21 Aug 2006 14:48:26 +0300 (EEST) Message-Id: <200608211148.k7LBmQGL161830@kosh.hut.fi> [ 28763] More conditional defines of EXTERN_C, by Jarkko [ 28765] Added an additional #ifndef based on comments by Jarkko and John Malmberg. [ 28767] Revert bogus change 28763. [ 28814] Subject: [PATCH] C++: Solaris CC now compiles "perl" From: Jarkko Hietaniemi Date: Sat, 09 Sep 2006 15:29:12 +0300 Message-ID: <4502B398.6060505@iki.fi> [ 28823] Subject: [PATCH] Re: C++ status [Fwd: Smoke [5.9.4] 28776 FAIL(FM) OSF1 V5.1 (EV6/4 cpu)] From: Dominic Dunlop Date: Tue, 12 Sep 2006 09:13:41 +0200 Message-Id: <6FD7503D-349B-4BE4-B457-D79458900FA0@computer.org> [ 28844] Subject: Re: Re: [PATCH] C++: Solaris CC now compiles "perl" From: demerphq Date: Thu, 14 Sep 2006 14:28:16 +0200 Message-ID: <9b18b3110609140528x49e063aay724169d7f1e49531@mail.gmail.com> [ 28848] More hacking in util.c's ifdef jungle following 28844 to avoid duplicate definition of Perl_my_setenv on VMS. [ 28860] Subject: [PATCH] util.c: restore Perl_my_setenv() to pre-28814 From: Jarkko Hietaniemi Date: Sun, 17 Sep 2006 10:29:08 +0300 Message-ID: <450CF944.7000402@iki.fi> Branch: maint-5.8/perl ! utf8.c !> EXTERN.h INTERN.h XSUB.h hints/solaris_2.sh mathoms.c op.h !> opcode.h opcode.pl perl.h perlio.c sv.c unixish.h util.c !> x2p/EXTERN.h x2p/INTERN.h x2p/hash.h ____________________________________________________________________________ [ 30033] By: nicholas on 2007/01/27 16:40:35 Log: Integrate: [ 28645] Subject: [PATCH] g++ heavy shoveling From: Jarkko Hietaniemi Date: Wed, 02 Aug 2006 00:04:42 +0300 Message-ID: <44CFC1EA.2050702@iki.fi> [ 28662] Subject: [PATCH] g++ large patch From: Jarkko Hietaniemi Date: Fri, 04 Aug 2006 08:58:27 +0300 Message-ID: <44D2E203.5050201@iki.fi> [ 28666] Restore modification in warnings.h from change 28662 by backporting it to warnings.pl [ 28674] Subject: [PATCH] g++ stage 1 reached From: Jarkko Hietaniemi Date: Tue, 08 Aug 2006 00:02:35 +0300 Message-ID: <44D7AA6B.4040802@iki.fi> Branch: maint-5.8/perl ! regcomp.c toke.c !> (integrate 45 files) ____________________________________________________________________________ [ 30031] By: nicholas on 2007/01/27 15:26:35 Log: Integrate: [ 28397] Subject: [PATCH] cflags.SH: rework the gcc warnings selection From: Jarkko Hietaniemi Date: Thu, 15 Jun 2006 09:12:53 +0300 Message-ID: <4490FA65.70006@iki.fi> [ 28402] Subject: RE: [PATCH] cflags.SH: rework the gcc warnings selection From: "Green, Paul" Date: Sun, 18 Jun 2006 19:41:50 -0400 Message-ID: [ 28434] Subject: [PATCH] cflags.SH: scan the gcc warning flags only once, during .SH expansion From: jhi@cc.hut.fi (Jarkko Hietaniemi) Date: Mon, 26 Jun 2006 22:12:37 +0300 (EEST) Message-Id: <20060626191237.8A0936CF04@seth.hut.fi> [ 28483] Don't add extra warning flags for gcc versions < 3. [ 28576] Subject: [PATCH] Re: [PATCH] cflags.SH: add -std=c89 From: Jarkko Hietaniemi Date: Sat, 15 Jul 2006 12:19:45 +0300 Message-ID: <44B8B331.8090009@iki.fi> [ 28577] Subject: [PATCH] cflags.SH: add missing C files From: Jarkko Hietaniemi Date: Sat, 15 Jul 2006 12:30:25 +0300 Message-ID: <44B8B5B1.2050902@iki.fi> [ 28592] Cygwin doesn't cope (yet) with gcc flags -std=c89 [ 28595] Remove the C89 checking with gcc (but leave a note in perlhack) Branch: maint-5.8/perl !> cflags.SH ____________________________________________________________________________ [ 30029] By: nicholas on 2007/01/27 13:10:53 Log: Integrate: [ 26135] Subject: [PATCH] Re: Cygwin status, maint and blead From: Yitzchak Scott-Thoennes Date: Sun, 13 Nov 2005 16:01:02 -0800 Message-ID: <20051114000102.GA5556@efn.org> [ 27004] Teach the Makefile how to cleanup config.over. [ 27214] Subject: Re: Smoke [5.9.4] 27195 FAIL(m) irix 6.2 (IP22/1 cpu) From: Abe Timmerman Message-Id: <200602172358.23977.abe@ztreet.demon.nl> Date: Fri, 17 Feb 2006 23:58:23 +0100 [ 27518] Simplify; make perl.o included in $(obj) [ 27519] Fix up miniperl dependency. Depend on $(obj) when we use that to build miniperl. [ 27891] When checking whether perl is built with -g in optimisation flags, use a more tolerant regexp, because we can now have things like '-O2 -g'. [ 28471] Subject: [PATCH] Makefile splint support From: andy@petdance.com (Andy Lester) Date: Sun, 2 Jul 2006 12:19:08 -0500 Message-ID: <20060702171908.GA20443@petdance.com> [ 29276] Subject: [PATCH] Makefile.SH: lib/re.pm often messes restarted parallel makes From: jhi@cc.hut.fi (Jarkko Hietaniemi) Date: Wed, 15 Nov 2006 05:48:32 +0200 (EET) Message-Id: <20061115034832.43B7E8B3E7@seth.hut.fi> [ 29497] Subject: Re: [PATCH 5.8.8] Build on OS/2 - missing part From: Ilya Zakharevich Date: Sat, 9 Dec 2006 16:33:41 -0800 Message-ID: <20061210003341.GA14200@powdermilk.math.berkeley.edu> Branch: maint-5.8/perl !> Makefile.SH os2/Makefile.SHs ____________________________________________________________________________ [ 30027] By: nicholas on 2007/01/27 12:18:00 Log: Integrate: [ 26993] Subject: Re: [patch] s/ar ts/ar s/g From: "Philippe M. Chiasson" Date: Thu, 26 Jan 2006 13:53:07 -0800 Message-ID: <43D944C3.60101@ActiveState.com> [ 28362] Subject: [PATCH]Teach Configure that icc is not gcc From: Steve Peters Date: Tue, 6 Jun 2006 10:33:12 -0500 Message-ID: <20060606153312.GA5542@fisharerojo.org> [ 28637] ANSI-C headers in test snippets to please g++ (and the rest) Subject: [PATCH] (back) towards g++ compilation (in Tru64, and elsewhere) From: Jarkko Hietaniemi Date: Mon, 31 Jul 2006 23:44:55 +0300 Message-ID: <44CE6BC7.8060805@iki.fi> [ 28651] Subject: [PATCH] g++ heavy shoveling From: Jarkko Hietaniemi Date: Wed, 02 Aug 2006 00:04:42 +0300 Message-ID: <44CFC1EA.2050702@iki.fi> Branch: maint-5.8/perl !> Configure ____________________________________________________________________________ [ 30026] By: nicholas on 2007/01/27 11:02:43 Log: Integrate: [ 28093] Subject: [PATCH] t/op/list.t using test.pl From: David Landgren Date: Wed, 03 May 2006 19:40:40 +0200 Message-ID: <4458EB18.8070904@landgren.net> [ 28649] Subject: [PATCH blead] Re: [perl #39733] $AUTOLOAD is never tainted From: Rick Delaney Date: Sun, 9 Jul 2006 15:01:50 -0400 Message-ID: <20060709190150.GA1922@localhost.localdomain> Plus a note in perldelta [ 28657] Subject: [PATCH blead] Re: [perl #39882] inconsistent list slice behaviour From: Rick Delaney Date: Thu, 3 Aug 2006 21:48:07 -0400 Message-ID: <20060804014807.GW21381@localhost.localdomain> [ 28684] Subject: Re: [perl #37731] junk and uninit'ed values in tied scalars From: Yitzchak Scott-Thoennes Date: Mon, 28 Nov 2005 01:26:31 -0800 Message-ID: <20051128092605.GA2328@efn.org> plus a regression test [ 28690] lstat on *GLOB{IO} with warnings on would segfault (noticed by Andrew Dougherty) [ 28827] Disable study() for utf-8 strings Subject: Re: Re: [Fwd: Smoke [5.9.4] 28821 FAIL(XF) OSF1 V5.1 (EV6/4 cpu)] From: demerphq Date: Tue, 12 Sep 2006 17:08:55 +0200 Message-ID: <9b18b3110609120808jceb84ces55ef29c2f2b73ebc@mail.gmail.com> [ 28875] Deliver SIGILL, SIGBUS and SIGSEGV always in an "unsafe" manner. Subject: Safe signals and SIGSEGV From: Rafael Garcia-Suarez Date: Fri, 8 Sep 2006 17:43:41 +0200 Message-ID: <20060908174341.0cc7988f@grubert.mandriva.com> Branch: maint-5.8/perl !> gv.c mg.c pp.c pp_sys.c sv.c t/op/list.t t/op/taint.t !> t/op/tie.t ____________________________________________________________________________ [ 30023] By: nicholas on 2007/01/26 21:52:35 Log: Integrate: [ 28603] In pp_entersub, accept to handle a CV with a NULL padlist. This fixes a crash with mod_perl 1.29. [ 28606] Add another volatile modifier to protect against longjmp clobbering [ 28607] Subject: [PATCH] z/OS: pp_sys.c, reg*.c, toke.c, utf8.c From: Jarkko Hietaniemi Date: Sat, 22 Jul 2006 18:51:48 +0300 Message-ID: <44C24994.6020008@iki.fi> [ 28608] Revert part of last change, that was causing utf8 test failures [ 28613] Revert change 28603: this was a wrong fix, that didn't prevent a later crash. Reindent a bit of the same function. [ 28634] Subject: [PATCH] solution(?) to the "warning: comparison is always false dueto limited range of data type" with New*() From: Jarkko Hietaniemi Date: Fri, 28 Jul 2006 09:00:28 +0300 Message-ID: <44C9A7FC.1060801@iki.fi> [ 28639] Subject: [PATCH] g++ dNOOP (luckily one can mix code and decls in C++) Somewhat modifies as per Steve Hay's comment From: Jarkko Hietaniemi Date: Tue, 1 Aug 2006 08:46:15 +0300 (EEST) Message-Id: <200608010546.k715kEdP017064@kosh.hut.fi> [ 28669] We shouldn't try to decrement the refcount of PL_warnhook if it has been set to PERL_WARNHOOK_FATAL. [ 28675] Simplification in Perl_magic_clearsig, plus a fix similar and symmetrical to the one implemented in change #28669. [ 28686] A couple const's and a cast to get Sun CC to compile these files. Perl, however, still will not build with the Sun CC. [ 28688] Subject: [PATCH] doio.c: vararg NULLs must be cast right From: Jarkko Hietaniemi Date: Thu, 10 Aug 2006 08:35:53 +0300 (EEST) Message-Id: <200608100535.k7A5Zr2Z156515@kosh.hut.fi> Branch: maint-5.8/perl !> doio.c handy.h mg.c op.c perl.h pp_hot.c pp_sort.c pp_sys.c !> regcomp.c regexec.c toke.c utf8.c ____________________________________________________________________________ [ 30022] By: nicholas on 2007/01/26 20:21:16 Log: Integrate: [ 28589] Subject: Re: Fix loads of warnings from the last escaping patch... From: demerphq Date: Sun, 16 Jul 2006 19:59:02 +0200 Message-ID: <9b18b3110607161059j276ac869p450aa178150044ae@mail.gmail.com> Branch: maint-5.8/perl !> dump.c embed.fnc pod/perlapi.pod proto.h ____________________________________________________________________________ [ 30021] By: nicholas on 2007/01/26 19:23:54 Log: Integrate: [ 28490] Introduce a new function, pv_escape(), to display contents of PVs that might contain non printable chars. Subject: Re: [PATCH]: fix: [perl #39583] Pattern Match fails for specific length string From: demerphq Date: Wed, 5 Jul 2006 20:40:58 +0200 Message-ID: <9b18b3110607051140n10c211a1jf17d3b7914d6f68b@mail.gmail.com> [ 28582] Subject: Updated escaping code. utf8 regex debug output improvements From: demerphq Date: Sat, 15 Jul 2006 18:56:03 +0200 Message-Id: <9b18b3110607150956o6273a16clb1518911d1945d4@mail.gmail.com> [ 28584] run regen_headerds Branch: maint-5.8/perl ! dump.c !> embed.fnc embed.h global.sym perl.h pod/perlapi.pod proto.h !> reentr.c ____________________________________________________________________________ [ 30020] By: nicholas on 2007/01/26 19:03:49 Log: Fix up files by re-running embed.pl I've no idea why these weren't correct to start with, as it all passed make test on saigo. Branch: maint-5.8/perl ! global.sym proto.h ____________________________________________________________________________ [ 30018] By: nicholas on 2007/01/26 17:39:35 Log: Integrate: [ 28512] Subject: [PATCH] Macroize vast tracks of duplicated code in regexec.c From: demerphq Date: Sat, 8 Jul 2006 17:48:34 +0200 Message-ID: <9b18b3110607080848u663b114ag86f76ac5c9c9d5e@mail.gmail.com> Removing duplicated regexp trie-related code. Branch: maint-5.8/perl ! regexec.c ____________________________________________________________________________ [ 30011] By: nicholas on 2007/01/26 14:31:27 Log: Integrate: [ 28370] Revert change #20462 (except the regression test), since the bug has been really fixed by change #22074, as explained in : Subject: Re: [perl #39247] defined-ness of substrings disappear over repeated calls From: SADAHIRO Tomoyuki Date: Thu, 01 Jun 2006 08:16:51 +0900 Message-Id: <20060601081304.7A39.BQW10602@nifty.com> [ 28548] Revert bogus fix for bug #27940, which wasn't really a bug, and a new test item, as found and suggested by Sadahiro Tomoyuki. [ 28620] Subject: interpolation of @- (and @+) in patterns ([perl #27940] comes back) From: SADAHIRO Tomoyuki Date: Tue, 25 Jul 2006 00:15:50 +0900 Message-Id: <20060725001517.3C5D.BQW10602@nifty.com> Branch: maint-5.8/perl !> pod/perlop.pod pp.c t/op/pat.t t/op/subst.t t/op/tr.t toke.c ____________________________________________________________________________ [ 30007] By: nicholas on 2007/01/26 13:50:55 Log: Integrate: [ 28525] Add Russ Allbery's public domain implementations of strlcat and strlcpy as Perl_my_strlcat and Perl_my_strlcpy to the Perl core. Thanks Russ! [ 28528] Change existing uses of strlcpy()/strlcat() to use new my_strlcpy()/ my_strlcat() API. Convert ext/File/Glob/bsd_glob.c to use my_strlcat(). Add to the strlcy()/strlcat() todo entry. [ 28533] Convert some low hanging fruit to my_strlcpy/my_strlcat. [ 28545] Fix change #28533: my_strlcpy does not return the dest str! Branch: maint-5.8/perl !> doio.c embed.fnc embed.h ext/File/Glob/bsd_glob.c global.sym !> mg.c perl.h pp_ctl.c pp_sys.c proto.h toke.c util.c ____________________________________________________________________________ [ 30006] By: nicholas on 2007/01/26 12:19:35 Log: Integrate: [ 27853] stop PL_op and PL_curpad referring to different pads at start of doeval (this could make find_uninit_var do Bad Things) [ 28523] Fix a signed/unsigned warning [ 28524] Add a macro to remove duplicated code Subject: Re: cut-and-paste findings for blead 28504 From: SADAHIRO Tomoyuki Date: Sun, 09 Jul 2006 13:47:04 +0900 Message-Id: <20060709134236.E51A.BQW10602@nifty.com> [ 28534] Silence empty if-statement warning. [ 28535] Fix another empty if-statement warning [ 28541] Subject: [PATCH] Re: [perl #39634] gcc 3.3 has problems with __attribute__((unused)) From: Andy Dougherty Date: Mon, 10 Jul 2006 13:16:27 -0400 (EDT) Message-ID: [ 28565] Clear up a few more warnings from blead. [ 28578] Subject: [PATCH] various safety/portability tweaks From: Jarkko Hietaniemi Date: Sat, 15 Jul 2006 13:14:32 +0300 Message-ID: <44B8C008.4030300@iki.fi> [ 28583] Subject: [PATCH] comment update for scan_const From: SADAHIRO Tomoyuki Date: Sat, 15 Jul 2006 20:16:01 +0900 Message-Id: <20060715201552.9FA5.BQW10602@nifty.com> [ 28585] [PATCH] Re: [PATCH] various safety/portability tweaks From: Jarkko Hietaniemi Date: Sat, 15 Jul 2006 22:15:29 +0300 Message-ID: <44B93ED1.7080000@iki.fi> Fix for change #28578. Branch: maint-5.8/perl !> doio.c op.c perl.c perl.h pp_ctl.c pp_sort.c pp_sys.c reentr.c !> reentr.pl sv.c toke.c universal.c util.c ____________________________________________________________________________ [ 30005] By: nicholas on 2007/01/26 12:01:34 Log: Remove spurious PERL_UNUSED_ARG()s that somehow don't cause errors round here, but make cause all sorts of problems elsewhere. Branch: maint-5.8/perl ! regcomp.c ____________________________________________________________________________ [ 30004] By: nicholas on 2007/01/26 11:57:08 Log: Integrate: [ 28468] Subject: [PATCH] Clean up mathoms.c From: andy@petdance.com (Andy Lester) Date: Sun, 2 Jul 2006 12:24:45 -0500 Message-ID: <20060702172445.GA20521@petdance.com> [ 28469] Subject: [PATCH] mg.c cleanups From: andy@petdance.com (Andy Lester) Date: Sun, 2 Jul 2006 12:33:36 -0500 Message-ID: <20060702173336.GA20641@petdance.com> [ 28477] Subject: [PATCH] consting and localizing in universal.c From: andy@petdance.com (Andy Lester) Date: Mon, 3 Jul 2006 17:41:48 -0500 Message-ID: <20060703224148.GA14449@petdance.com> Branch: maint-5.8/perl ! mathoms.c !> embed.fnc mg.c proto.h universal.c ____________________________________________________________________________ [ 30003] By: nicholas on 2007/01/26 11:36:22 Log: Integrate: [ 28408] Subject: [PATCH] Re: [perl #39530] printf: bad formatting of hexadecimal conversion of binary string using vector flag From: Dominic Dunlop Date: Tue, 20 Jun 2006 16:02:38 +0200 Message-Id: [ 28411] Subject: [PATCH] Re: [PATCH] Re: [perl #39530] printf: bad formatting of hexadecimal conversion of binary string using vector flag From: Dominic Dunlop Date: Tue, 20 Jun 2006 21:49:07 +0200 Message-Id: Branch: maint-5.8/perl !> sv.c t/op/sprintf.t ____________________________________________________________________________ [ 29999] By: nicholas on 2007/01/26 11:02:10 Log: Integrate: [ 28327] Silence a warning [ 28413] Subject: [PATCH] isALNUM(*s) && *s != '_' From: "Dr.Ruud" Date: Wed, 21 Jun 2006 01:58:19 +0200 Message-ID: <20060621000237.24398.qmail@lists.develooper.com> [ 28447] Some CPP macro sanitization by Sadahiro Tomoyuki [ 28453] HASATTRIBUTE_UNUSED works for gcc-3.3, but not for g++-3.3 Subject: Re: [PATCH] Re: [perl #39634] gcc 3.3 has problems with __attribute__((unused)) From: Andrew Dougherty Date: Thu, 29 Jun 2006 13:41:28 -0400 (EDT) Message-ID: [ 28456] Fix a couple of compilation warnings [ 28457] PERL_UNUSED_CONTEXT is not enough to silence warnings when pTHXx is used. [ 28459] Fix casting warnings related to snprintf/vsnprintf [ 28465] remove some unnecessary uses of WITH_THR [ 28476] Subject: [PATCH] toke.c consting and localizing From: andy@petdance.com (Andy Lester) Date: Mon, 3 Jul 2006 17:42:50 -0500 Message-ID: <20060703224250.GA14481@petdance.com> Branch: maint-5.8/perl !> mg.c op.c perl.c perl.h regcomp.c toke.c utfebcdic.h util.c ____________________________________________________________________________ [ 29997] By: nicholas on 2007/01/26 10:30:23 Log: Integrate: [ 28381] Subject: [PATCH] Proper use of enums From: andy@petdance.com (Andy Lester) Date: Sun, 11 Jun 2006 22:44:34 -0500 Message-ID: <20060612034434.GA21588@petdance.com> [ 28385] Subject: [PATCH] Static cleanup in pp_sort.c From: andy@petdance.com (Andy Lester) Date: Mon, 12 Jun 2006 08:59:34 -0500 Message-ID: <20060612135934.GA2202@petdance.com> Branch: maint-5.8/perl !> embed.fnc embed.h gv.c op.c perlio.c pod/perlapi.pod pp.c !> pp_sort.c proto.h sv.c sv.h ____________________________________________________________________________ [ 29996] By: nicholas on 2007/01/26 09:54:13 Log: Integrate: [ 28376] Subject: [PATCH] Add error checks after execing PL_cshname or PL_sh_path From: "Jan Dubois" Date: Wed, 7 Jun 2006 15:53:02 -0700 Message-ID: <002301c68a85$21aa7320$d045a8c0@candy> [ 28377] Mark the 1st argument to S_exec_failed as non null [ 28404] Subject: [perl #39365] Bug in toke.c (eval in subst) From: bcarter@gumdrop.flyinganvil.org (via RT) Date: Fri, 09 Jun 2006 02:08:44 -0700 Message-ID: plus a regression test. [ 28409] Change 28404 broke the construct s/foo/< doio.c embed.fnc embed.h proto.h t/comp/parser.t toke.c ____________________________________________________________________________ [ 29995] By: nicholas on 2007/01/26 09:31:24 Log: Integrate: [ 28328] Subject: Re: [perl #39126] possible memory related bug when using sprintf with an utf-8 encoded format-string and iso-8859-1 encoded string variables. From: SADAHIRO Tomoyuki Date: Sun, 21 May 2006 19:33:21 +0900 Message-Id: <20060521193259.81E5.BQW10602@nifty.com> with test tweaks further suggested by the same [ 28331] make require report too many open files error ie don't continue searching the rest of @INC [ 28364] Subject: [PATCH] exhausting <> in BEGIN{} gets ARGVOUT used only once warning From: Yitzchak Scott-Thoennes Date: Tue, 6 Jun 2006 22:24:46 -0700 Message-ID: <20060607052446.GA3240@efn.org> Branch: maint-5.8/perl !> gv.c pp_ctl.c sv.c t/op/readline.t t/op/sprintf2.t ____________________________________________________________________________ [ 29993] By: nicholas on 2007/01/26 09:15:17 Log: Integrate: [ 28279] Subject: [PATCH] pp_sys cleanup again From: andy@petdance.com (Andy Lester) Date: Sat, 20 May 2006 16:20:20 -0500 Message-ID: <20060520212020.GA30688@petdance.com> [ 28289] Subject: [PATCH] clean some lint findings From: andy@petdance.com (Andy Lester) Date: Sun, 21 May 2006 00:02:40 -0500 Message-ID: <20060521050240.GA21051@petdance.com> [ 28296] Subject: [PATCH] Refactoring in av.c From: andy@petdance.com (Andy Lester) Date: Wed, 24 May 2006 00:16:40 -0500 Message-ID: <20060524051640.GA4713@petdance.com> [ 28357] Subject: [PATCH] Accumulated toke.c cleanups From: andy@petdance.com (Andy Lester) Date: Tue, 6 Jun 2006 00:07:36 -0500 Message-ID: <20060606050736.GA30291@petdance.com> [ 28358] Subject: [PATCH] more accumulated cleanups From: andy@petdance.com (Andy Lester) Date: Tue, 6 Jun 2006 00:25:01 -0500 Message-ID: <20060606052501.GA30469@petdance.com> [ 28363] Subject: [PATCH] dump.c patches From: andy@petdance.com (Andy Lester) Date: Tue, 6 Jun 2006 10:01:37 -0500 Message-ID: <20060606150137.GA4434@petdance.com> Branch: maint-5.8/perl ! util.c !> av.c cop.h doio.c doop.c dump.c embed.fnc embed.h hv.c !> perlio.c pp_pack.c pp_sys.c proto.h toke.c utf8.c ____________________________________________________________________________ [ 29988] By: nicholas on 2007/01/25 23:24:46 Log: Integrate: [ 28218] Add a new macro SvVSTRING_mg that returns vstring magic, if any. SvVOK used to do that, in stealth mode, but now it returns a boolean like it used to. Patch suggested by John Peacock. Branch: maint-5.8/perl !> sv.c sv.h ____________________________________________________________________________ [ 29986] By: nicholas on 2007/01/25 23:07:41 Log: Integrate: [ 28251] [perl #32332] Perl segfaults; test case available sub f { s/$var/f()/e } could free the wrong RE [ 28254] eval { s/$foo/die/e } leaked a scalar Branch: maint-5.8/perl !> cop.h pp_ctl.c pp_hot.c ____________________________________________________________________________ [ 29985] By: nicholas on 2007/01/25 22:55:28 Log: Integrate: [ 28193] Subject: [PATCH] strange encodings upsets pp_chr From: SADAHIRO Tomoyuki Date: Sun, 14 May 2006 19:57:28 +0900 Message-Id: <20060514195532.5422.BQW10602@nifty.com> [ 28215] Subject: Re: [perl #39145] win32, @_ and fork crashing in dounwind From: Dave Mitchell Date: Wed, 17 May 2006 17:38:16 +0100 Message-ID: <20060517163816.GN17109@iabyn.com> (Dave's fix, plus a test taken from [perl #39145]) [ 28248] [perl #32041] SEGV with complicated regexp and long string PL_reg_maxiter was wrapping to a negative value Branch: maint-5.8/perl +> t/uni/chr.t !> MANIFEST op.c pp.c regexec.c sv.c t/op/chr.t t/op/fork.t !> t/op/ord.t ____________________________________________________________________________ [ 29984] By: nicholas on 2007/01/25 22:41:11 Log: Integrate: [ 28183] Subject: Re: [PATCH] my_snprintf From: Jarkko Hietaniemi Date: Fri, 12 May 2006 22:28:49 +0300 Message-ID: <4464E1F1.9010706@gmail.com> [ 28186] Subject: [PATCH] sv.c: printf %d wants int not size_t From: Jarkko Hietaniemi Message-Id: <200605130715.k4D7FWg3089581@kosh.hut.fi> Date: Sat, 13 May 2006 10:15:32 +0300 (EEST) [ 28216] Subject: Re: Change 28183 has broken 64-bit builds? From: Jarkko Hietaniemi Date: Wed, 17 May 2006 22:19:51 +0300 Message-ID: <446B7757.8030301@gmail.com> [ 28217] Parenthesize uses of macro arguments [ 28222] Adjust calling of Perl_va_copy(), noticed by Jarkko [ 28234] Subject: [PATCH] Re: [PATCH] Re: Change 28183 has broken 64-bit builds? From: Jarkko Hietaniemi Date: Thu, 18 May 2006 22:44:14 +0300 Message-ID: <446CCE8E.8010407@gmail.com> [ 28247] Perl_croak() needs an aTHX_ in PerlIO_vsprintf(). [ 28249] Solaris was happy, but change 28247 removed the wrong dTHX as far as some other operating systems were concerned. [ 28299] The 'f' flag was missing for my_snprintf(). Branch: maint-5.8/perl !> embed.fnc global.sym perl.h perlio.c pod/perlapi.pod pp_ctl.c !> proto.h regcomp.c sv.c toke.c universal.c util.c ____________________________________________________________________________ [ 29982] By: nicholas on 2007/01/25 22:04:51 Log: Integrate: [ 28132] Subject: [PATCH] PERL_MEM_LOG enhancements From: Jarkko Hietaniemi Date: Sat, 06 May 2006 11:21:02 +0300 Message-ID: <445C5C6E.6070201@gmail.com> Branch: maint-5.8/perl !> util.c ____________________________________________________________________________ [ 29981] By: nicholas on 2007/01/25 21:31:37 Log: Integrate: [ 28205] Subject: [PATCH] refactoring out S_regoptail From: andy@petdance.com (Andy Lester) Date: Sun, 14 May 2006 23:15:18 -0500 Message-ID: <20060515041518.GA15745@petdance.com> [ 28206] Subject: [PATCH] Refactoring reghops From: andy@petdance.com (Andy Lester) Date: Mon, 15 May 2006 08:54:24 -0500 Message-ID: <20060515135424.GA31585@petdance.com> [ 28209] Subject: Re: [PATCH] upgrade bytes_to_uni From: SADAHIRO Tomoyuki Date: Mon, 15 May 2006 22:39:58 +0900 Message-Id: <20060515223901.F3B0.BQW10602@nifty.com> [ 28236] strlen(foo) rather than strchr(foo, 0) makes simpler code, and is likely to be a more efficient implementation. [ 28237] Subject: [PATCH] strlcat aaargh From: Jarkko Hietaniemi Date: Fri, 19 May 2006 20:44:46 +0300 Message-ID: <446E040E.8030205@gmail.com> [ 28246] There is no STASH type. Branch: maint-5.8/perl !> doio.c embed.fnc embed.h perlio.c pp_pack.c proto.h regcomp.c !> regexec.c sv.h ____________________________________________________________________________ [ 29979] By: nicholas on 2007/01/25 21:01:25 Log: Re-run embed.pl. Branch: maint-5.8/perl ! global.sym proto.h ____________________________________________________________________________ [ 29978] By: nicholas on 2007/01/25 21:01:04 Log: Fix 29976 (somehow I can't have run make test) - in maint there need to be push_return() and pop_return() wrapping the faked up eval for constant folding. Branch: maint-5.8/perl ! op.c ____________________________________________________________________________ [ 29976] By: nicholas on 2007/01/25 17:25:09 Log: Integrate: [ 28135] allow exit during fold_constants eg BEGIN { $SIG{__WARN__} = sub{exit};} "a" == "b" [ 28148] disable WARN and DIE hooks during constant folding Branch: maint-5.8/perl !> op.c t/comp/fold.t util.c warnings.h warnings.pl ____________________________________________________________________________ [ 29974] By: nicholas on 2007/01/25 17:04:16 Log: Integrate: [ 28069] Add a new hash key flag HVhek_UNSHARED, to mark all unshared hash keys. Otherwise Perl_newSVhek(), called by Perl_hv_iterkeysv(), naively assumes that it can share any hash key directly. [ 28074] Subject: [PATCH] microperl update From: Jarkko Hietaniemi Date: Wed, 03 May 2006 08:20:31 +0300 Message-ID: <44583D9F.8060401@gmail.com> [ 28075] Subject: [PATCH] hv.c: do zeroing via mem wrap check From: Jarkko Hietaniemi Date: Tue, 2 May 2006 20:03:01 +0300 (EEST) Message-Id: <200605021703.k42H31PF216868@kosh.hut.fi> [ 28076] Subject: [PATCH] pp_sys.c: U8* cast From: Jarkko Hietaniemi Date: Tue, 2 May 2006 20:04:33 +0300 (EEST) Message-Id: <200605021704.k42H4XIh255652@kosh.hut.fi> [ 28077] Subject: [PATCH] clean up loops in doio.c and dump.c From: andy@petdance.com (Andy Lester) Date: Tue, 2 May 2006 13:38:15 -0500 Message-ID: <20060502183815.GA7979@petdance.com> [ 28079] Subject: [PATCH] Remove an unnecessary goto From: andy@petdance.com (Andy Lester) Date: Wed, 3 May 2006 00:54:30 -0500 Message-ID: <20060503055430.GA4922@petdance.com> [ 28081] Subject: [PATCH] more logic localizations From: andy@petdance.com (Andy Lester) Date: Tue, 11 Apr 2006 14:14:45 -0500 Message-ID: <20060411191445.GA4938@petdance.com> [ 28136] Fix a logic error in conditional found by Coverity [ 28137] Fix another Coverity problem by adding an assert that says that some string we got from the lexer is null-terminated Branch: maint-5.8/perl ! hv.c op.c !> Makefile.micro README.micro doio.c dump.c embed.fnc embed.h !> hv.h mg.c pp_sys.c proto.h sv.c uconfig.h uconfig.sh ____________________________________________________________________________ [ 29971] By: nicholas on 2007/01/25 11:59:28 Log: Integrate: [ 28058] De-duplicate the code that creates new GPs into Perl_newGP(). [back out the gv.c part of:] [ 29735] We can get in the messy situation of the COP that PL_curcop pointed to getting freed, and as part of the same free overloading decides to look for DESTROY, which needs to *create* a GV, which in turn was expecting that PL_curcop pointed to something valid. So set PL_curcop to NULL if we're freeing the COP that it points to, and make Perl_gv_init() cope with a NULL PL_curcop. [ 29748] An implementation of change 29735 for blead (PL_curcop could be NULL) given that blead's refactoring is not yet in maint. [ 29970] Change 28058 hadn't been tested with -DPERL_CREATE_GVSV Branch: maint-5.8/perl ! gv.c !> embed.fnc embed.h proto.h scope.c ____________________________________________________________________________ [ 29969] By: nicholas on 2007/01/25 10:55:58 Log: Go away, stowaway file. Branch: maint-5.8/perl - ext/DynaLoader/dl_symbian.xs ____________________________________________________________________________ [ 29968] By: nicholas on 2007/01/25 10:55:13 Log: Integrate: [ 28063] Fix bug in DynaLoader, which has been passing a filename in dynamic storage to newXS() seemingly forever. This involves creating newXS_flags(), with the first flag being "arrange to copy the filename and free it at the right time". [ 28064] Change 28063 forgot 1 key line - acutally use the correct filename storage. [ 28066] *static* pointers to newXSUB(), dammit! [ 29107] Plug a memory leak in newCONSTSUB Branch: maint-5.8/perl +> ext/DynaLoader/dl_symbian.xs !> XSUB.h cv.h embed.fnc embed.h ext/DynaLoader/dl_aix.xs !> ext/DynaLoader/dl_beos.xs ext/DynaLoader/dl_dld.xs !> ext/DynaLoader/dl_dllload.xs ext/DynaLoader/dl_dlopen.xs !> ext/DynaLoader/dl_dyld.xs ext/DynaLoader/dl_hpux.xs !> ext/DynaLoader/dl_mac.xs ext/DynaLoader/dl_mpeix.xs !> ext/DynaLoader/dl_next.xs ext/DynaLoader/dl_vmesa.xs !> ext/DynaLoader/dl_vms.xs op.c pod/perlapi.pod proto.h !> writemain.SH xsutils.c ____________________________________________________________________________ [ 29967] By: nicholas on 2007/01/25 10:34:50 Log: Integrate: [ 28047] Subject: [PATCH] More tweaking of SvREFCNT_inc calls From: andy@petdance.com (Andy Lester) Date: Sun, 30 Apr 2006 12:06:21 -0500 Message-ID: <20060430170621.GA4054@petdance.com> [ 28048] Subject: [PATCH] quiet some perlio.c warnings From: andy@petdance.com (Andy Lester) Date: Sun, 30 Apr 2006 13:23:22 -0500 Message-ID: <20060430182322.GA18064@petdance.com> [ 28049] Subject: [PATCH] Some more SvREFCNT_inc updates From: andy@petdance.com (Andy Lester) Date: Sun, 30 Apr 2006 13:24:14 -0500 Message-ID: <20060430182414.GA18094@petdance.com> [ 28050] Subject: [PATCH] pp.c casting, take two From: andy@petdance.com (Andy Lester) Date: Sun, 30 Apr 2006 19:48:26 -0500 Message-ID: <20060501004826.GA4744@petdance.com> (with adjustements, some of the code has already been touched since) Branch: maint-5.8/perl !> cop.h mg.c perlio.c pp.c pp_ctl.c ____________________________________________________________________________ [ 29966] By: nicholas on 2007/01/25 10:15:27 Log: Correct the definition of S_vdie_common in mathoms.c Branch: maint-5.8/perl ! mathoms.c ____________________________________________________________________________ [ 29965] By: nicholas on 2007/01/25 00:07:28 Log: Integrate: [ 28031] Perl_pack_cat() is a mathom too! Branch: maint-5.8/perl !> mathoms.c pod/perlapi.pod pp_pack.c ____________________________________________________________________________ [ 29964] By: nicholas on 2007/01/24 23:53:28 Log: Integrate: [ 28008] Subject: [PATCH] reduce gcc -ansi -pedantic noise plus a suggestion From: Jarkko Hietaniemi Date: Fri, 28 Apr 2006 22:58:58 +0300 Message-ID: <44527402.8000506@gmail.com> [ 28010] Subject: [PATCH] more -ansi -pedantic cleanliness From: Jarkko Hietaniemi Date: Sat, 29 Apr 2006 11:13:24 +0300 Message-ID: <44532024.9070303@gmail.com> [ 28405] Clean up a few warnings in regcomp.c under -Dgccansipedantic. [ 28412] Fix code before declaration error introduced by #28405 Branch: maint-5.8/perl ! scope.h sv.c !> cop.h doio.c gv.c hv.c op.c pad.c perl.c perl.h pp.c pp_ctl.c !> pp_hot.c pp_pack.c pp_sort.c pp_sys.c regcomp.c regexec.c !> toke.c universal.c utf8.c ____________________________________________________________________________ [ 29963] By: nicholas on 2007/01/24 23:24:03 Log: Integrate: [ 27976] Subject: [PATCH] Removing unused context From: andy@petdance.com (Andy Lester) Date: Thu, 27 Apr 2006 00:32:15 -0500 Message-ID: <20060427053215.GA31445@petdance.com> [ 27992] Subject: [PATCH] Removing another context parm From: andy@petdance.com (Andy Lester) Date: Thu, 27 Apr 2006 10:31:56 -0500 Message-ID: <20060427153156.GA11445@petdance.com> [ 28020] Coverity disliked us passing a negative length to vsnprintf(). [ 28023] remove a compiler warning by making HOPBACKc only hop *back* [ 28032] Replace a STRLEN no_len; with a _nolen macro call. [ 28041] Reverting context removal from change #27992. This should clear up threaded builds on Win32 and VMS. [ 28055] Subject: [PATCH] toke.c: function pointer to data pointer From: Jarkko Hietaniemi Date: Tue, 2 May 2006 08:25:55 +0300 (EEST) Message-Id: <200605020525.k425Ptik165705@kosh.hut.fi> Branch: maint-5.8/perl ! proto.h !> embed.fnc embed.h perlio.c pp_ctl.c pp_pack.c regexec.c sv.c !> toke.c ____________________________________________________________________________ [ 29962] By: nicholas on 2007/01/24 22:51:14 Log: Integrate: [ 27962] Subject: Re: [PATCH] cleanup 212 warnings emitted by gcc-4.2 From: Marcus Holland-Moritz Date: Mon, 24 Apr 2006 23:20:38 +0200 Message-ID: <20060424232038.7550f9b6@r2d2> Branch: maint-5.8/perl !> deb.c doio.c doop.c dump.c gv.c handy.h mg.c op.c patchlevel.h !> perl.c perlio.c pp.c pp_ctl.c pp_hot.c pp_sys.c regcomp.c !> regexec.c scope.h sv.c toke.c util.c ____________________________________________________________________________ [ 29961] By: nicholas on 2007/01/24 22:12:24 Log: Integrate: [ 27948] Subject: [perl #38475] attribute multiline fix (in tokenizer) From: bas@quarantainenet.nl (via RT) Date: Thu, 09 Feb 2006 06:33:51 -0800 Message-ID: [ 27950] Add regression test for bug #38475 [ 27986] [perl #39012] another REIFY bug Branch: maint-5.8/perl +> t/op/attrhand.t !> MANIFEST scope.c t/op/local.t toke.c ____________________________________________________________________________ [ 29959] By: nicholas on 2007/01/24 20:33:04 Log: Integrate: [ 27945] A couple of casts to PADOFFSET. 2nd patch from : Subject: Re: [PATCH] cleanup 212 warnings emitted by gcc-4.2 From: Marcus Holland-Moritz Date: Sun, 23 Apr 2006 04:47:04 +0200 Message-ID: <20060423044704.6a383ee8@r2d2> [ 27959] Coverity has started to notice another inconsistency in our NULL pointer checking. Branch: maint-5.8/perl !> op.c pad.c pp.c pp_ctl.c toke.c ____________________________________________________________________________ [ 29958] By: nicholas on 2007/01/24 19:49:21 Log: Integrate: [ 27943] Subject: [PATCH] doop.c consting, take 2 From: andy@petdance.com (Andy Lester) Date: Sun, 23 Apr 2006 20:45:09 -0500 Message-ID: <20060424014509.GA29642@petdance.com> [ 27958] Subject: [PATCH] use NOOP macro From: andy@petdance.com (Andy Lester) Date: Mon, 24 Apr 2006 13:44:51 -0500 Message-ID: <20060424184451.GA1479@petdance.com> Branch: maint-5.8/perl !> doop.c embed.fnc gv.c gv.h hv.c op.c pad.c perl.h perlio.c !> pp.c pp_ctl.c pp_hot.c proto.h regcomp.c regexec.c sv.c toke.c !> utf8.c ____________________________________________________________________________ [ 29957] By: nicholas on 2007/01/24 19:17:41 Log: Integrate: [ 27909] Subject: [PATCH] regexec consting & localizing From: andy@petdance.com (Andy Lester) Date: Wed, 19 Apr 2006 12:35:43 -0500 Message-ID: <20060419173543.GA8268@petdance.com> [ 27911] regmatch(): replace magic value -1000 with symbolic CHRTEST_VOID Branch: maint-5.8/perl !> regexec.c ____________________________________________________________________________ [ 29955] By: nicholas on 2007/01/24 18:58:36 Log: Integrate: [ 27896] Calling cv_undef() on the CV created by newCONSTSUB() would leak like a Jumblie's preferred maritime craft. To free CvFILE for this case, take advantage of the 0 length prototype that will also be there, and hang it from the prototype. To do this properly means changing code to actually pay attention to SvCUR() on prototypes. It turns out that we always know the length of the prototype string, so this may be faster. Certainly, it's a memory saving (even ignoring the leak). [ 27898] Avoid temporarily writing over the prototype when reporting an error. (And beef up the relevant tests to really check that it all works). [ 27900] Oops. Need a macro to convert cv_ckproto() to cv_ckproto_len(). [ 27901] newCONSTSUB needs to be robust in case CopFILE is NULL. [ 29954] Wrap the macro arguments for ck_proto in (). Branch: maint-5.8/perl ! op.h !> dump.c embed.fnc embed.h global.sym mathoms.c op.c !> pod/perlapi.pod proto.h sv.c t/comp/proto.t util.c ____________________________________________________________________________ [ 29953] By: nicholas on 2007/01/24 17:01:34 Log: Integrate: [ 27792] Remove SAVEt_FREESHAREDPV, as nothing is using it, and it isn't catered for in Perl_ss_dup, hence nothing is testing it and nothing will alert us if it breaks. [ 27866] Add Dave's explainations of why certain pointers are always non-NULL, and assert() that it is true. [ 27869] Subject: [PATCH] (tentative) regcomp.c: try convincing Coverity that data ptr in study_chunk is non-NULL when needed From: jhi@cc.hut.fi (Jarkko Hietaniemi) Message-Id: <20060416092648.812D36CF2D@aprikoosi.hut.fi> Date: Sun, 16 Apr 2006 12:26:48 +0300 (EEST) [ 27894] Revert 27869 at Jarkko's request. Branch: maint-5.8/perl ! regexec.c !> regcomp.c scope.c scope.h ____________________________________________________________________________ [ 29952] By: nicholas on 2007/01/24 16:46:32 Log: Integrate: [ 27871] Subject: Re: [PATCH] update to pp_complement() via Coverity From: SADAHIRO Tomoyuki Message-Id: <20060416184044.4ADB.BQW10602@nifty.com> Date: Sun, 16 Apr 2006 18:45:37 +0900 Subject: Re: [PATCH] update to pp_complement() via Coverity From: SADAHIRO Tomoyuki Message-Id: <20060417230614.20A5.BQW10602@nifty.com> Date: Mon, 17 Apr 2006 23:06:21 +0900 [ 27877] Coverity is upset about a signed int in ext/Filter/Util/Call/Call.xs, but the real problem is down in Perl_filter_read() in toke.c. [ 27882] Coverity insists that if we deference a pointer without checking, and subsequently check that same pointer before deferencing it, something in our code or our logic is bogus. So assert() that it's safe the first time and remove the second check. [ 27895] Fix compilation of microperl, which doesn't have INT_MAX. [ 27897] Replace some Copy() by Move() calls, because valgrind reports we can have overlapping memory areas here [ 27908] Subject: [PATCH] Localizing vars From: andy@petdance.com (Andy Lester) Date: Wed, 19 Apr 2006 23:44:01 -0500 Message-ID: <20060420044401.GA2792@petdance.com> [ 27910] Subject: [PATCH] op_type shrinking From: andy@petdance.com (Andy Lester) Date: Wed, 19 Apr 2006 23:16:53 -0500 Message-ID: <20060420041653.GA28523@petdance.com> Branch: maint-5.8/perl !> doop.c dump.c pp.c regcomp.c sv.c t/op/bop.t toke.c util.c ____________________________________________________________________________ [ 29951] By: nicholas on 2007/01/24 16:24:04 Log: Integrate: [ 27840] Merge sv_usepvn and sv_usepvn_mg into sv_usepvn_flags. "Promote" the other two to mathoms.c [ 27841] Provide a new flag for sv_usepvn_flags, SV_HAS_TRAILING_NUL, which signals that the buffer passed in is already well-formed for use as SvPVX (i.e. it is followed by a trailing NUL), and hence the realloc() to add a NUL can be skipped. [ 27842] In Perl_sv_usepvn_flags, assert() that the caller has RTFM. [ 27843] Perl_do_vop can profitably use Perl_sv_usepvn_flags, as it has allocated space for an initialised a trailing NUL. [ 27846] Stray pod/perlapi.pod change found after a regen. [ 27852] Fix bug introduced with change 27841. (len + 1, not just len). Branch: maint-5.8/perl !> doop.c embed.fnc embed.h global.sym mathoms.c pod/perlapi.pod !> pp.c proto.h sv.c sv.h ____________________________________________________________________________ [ 29950] By: nicholas on 2007/01/24 15:42:09 Log: Integrate: [ 27801] Subject: [PATCH] doop.c: (Coverity) found a bug but not quite what Coverity thought it did (try valgrind on the new bop.t without the doop.c patch) From: jhi@cc.hut.fi (Jarkko Hietaniemi) Date: Thu, 13 Apr 2006 19:20:46 +0300 (EEST) Message-Id: <20060413162046.5F9636D08C@ugli.hut.fi> [ 27856] The danger of piping an mbox to patch is that it contains more than one message. So both: Subject: [PATCH] doop.c: one more code path where memory could leak (Coverity) From: jhi@cc.hut.fi (Jarkko Hietaniemi) Message-Id: <20060416081925.680336CF2D@aprikoosi.hut.fi> Date: Sun, 16 Apr 2006 11:19:25 +0300 (EEST) and Subject: [PATCH] doop.c: one more code path where memory could leak (Coverity) From: jhi@cc.hut.fi (Jarkko Hietaniemi) Message-Id: <20060416081925.680336CF2D@aprikoosi.hut.fi> Date: Sun, 16 Apr 2006 11:19:25 +0300 (EEST) [ 27857] Revert 27856. [ 27859] Subject: [PATCH] dooop.c: the strong asserts in Sv* macros could cause memory leakage -- move the macro calls earlier (Coverity CID 84) From: jhi@cc.hut.fi (Jarkko Hietaniemi) Message-Id: <20060417071937.C13346CF2D@aprikoosi.hut.fi> Date: Mon, 17 Apr 2006 10:19:37 +0300 (EEST) [ 27883] Coverity still thinks that there is a route through do_vop that can leak resources. I believe that it's spotted that you can skip all the cases in the switch. Plug that hole. Branch: maint-5.8/perl !> doop.c pod/perlapi.pod sv.c t/op/bop.t ____________________________________________________________________________ [ 29949] By: nicholas on 2007/01/24 15:28:38 Log: Integrate: [ 27641] Subject: [PATCH] Random accumulated patches From: andy@petdance.com (Andy Lester) Date: Thu, 30 Mar 2006 23:42:28 -0600 Message-ID: <20060331054228.GA18940@petdance.com> [somehow the perl.h patch got missed] [ 27754] Subject: [PATCH]es: quiet some warnings (with a free AIX cpp insanity avoidance tip) From: Jarkko Hietaniemi Message-ID: <44375F23.6030900@gmail.com> Date: Sat, 08 Apr 2006 09:58:43 +0300 [ 27763] Subject: [PATCH] Quiet a coverity problem From: andy@petdance.com (Andy Lester) Date: Mon, 10 Apr 2006 23:51:50 -0500 Message-ID: <20060411045150.GA1422@petdance.com> [ 27766] Stop Perl_newSVpvn_share() potentially leaking the return result from bytes_from_utf8(). [ 27769] Subject: [PATCH] locale.c: more Safefree() (Coverity finding) From: Jarkko Hietaniemi Message-Id: <200604111908.k3BJ8ewn030950@kosh.hut.fi> Date: Tue, 11 Apr 2006 22:08:40 +0300 (EEST) [ 27770] Subject: [PATCH] perlio.c: false Coverity finding (it cannot see that two pointers are the same): but help Coverity by removing the temp ptr From: Jarkko Hietaniemi Message-Id: <200604112039.k3BKdWix334525@kosh.hut.fi> Date: Tue, 11 Apr 2006 23:39:32 +0300 (EEST) [ 27771] Subject: [PATCH] regcomp.c (try II): use &PL_sv_undef for listsv as suggested by Nick From: Jarkko Hietaniemi Message-Id: <200604112053.k3BKrPgk310001@kosh.hut.fi> Date: Tue, 11 Apr 2006 23:53:25 +0300 (EEST) [ 27772] Subject: [PATCH] regexec.c: move NULL check earlier to make Coverity happier From: Jarkko Hietaniemi Message-Id: <200604112004.k3BK4mV3097400@kosh.hut.fi> Date: Tue, 11 Apr 2006 23:04:48 +0300 (EEST) [ 27829] Subject: [PATCH] Handle null io in leavewrite (not possible?) (Coverity) From: bah@longitude.com Message-ID: <1661.38.96.138.153.1145124821.squirrel@mail.longitude.com> Date: Sat, 15 Apr 2006 14:13:41 -0400 (EDT) [ 27836] Subject: [PATCH] update to pp_complement() via Coverity From: andy@petdance.com (Andy Lester) Date: Sun, 16 Apr 2006 00:14:23 -0500 Message-ID: <20060416051423.GA17063@petdance.com> [ 27837] Coverity notes that we might be dereferencing fgv before a check that it's not NULL. In fact, the code ordering meant that one "if (fgv)" would always be true. So fix this. [ 27867] Add assertions to cover cases where the Coverity scanner thinks we could are deferencing a NULL pointer, but we think it missed something it should have been capable of figuring out. [ 27868] Note in av_store() that S_regclass relies on being able to pass NULL. Branch: maint-5.8/perl !> av.c hv.c locale.c op.c pad.h perl.h perlio.c pp.c pp_sys.c !> regcomp.c regexec.c sv.c ____________________________________________________________________________ [ 29948] By: nicholas on 2007/01/24 15:13:14 Log: Integrate: [ 27753] Subject: [PATCH] True consting goodness in regexes From: andy@petdance.com (Andy Lester) Date: Mon, 10 Apr 2006 02:15:58 -0500 Message-Id: <20060410071558.GA24756@petdance.com> [ 27762] All S_dumpuntil()'s regnode pointer arguments can be const. Branch: maint-5.8/perl ! proto.h regcomp.c !> embed.fnc regexec.c ____________________________________________________________________________ [ 29947] By: nicholas on 2007/01/24 13:54:09 Log: Integrate: [ 27749] Subject: [PATCH] Rename variables in some mg_* funcs From: andy@petdance.com (Andy Lester) Date: Sun, 9 Apr 2006 00:20:12 -0500 Message-ID: <20060409052012.GA15887@petdance.com> [ 27834] Subject: [PATCH] Removed unused var in a macro From: andy@petdance.com (Andy Lester) Date: Sat, 15 Apr 2006 23:32:53 -0500 Message-ID: <20060416043253.GA3438@petdance.com> [ 27835] Subject: [PATCH] clean up hv_assert() From: andy@petdance.com (Andy Lester) Date: Sat, 15 Apr 2006 23:18:27 -0500 Message-ID: <20060416041827.GA3031@petdance.com> [ 27845] Subject: [PATCH] dump.c patch redux From: andy@petdance.com (Andy Lester) Date: Sun, 16 Apr 2006 00:29:36 -0500 Message-ID: <20060416052936.GA19143@petdance.com> [ 27850] Subject: [PATCH] op.c patch, 2nd version From: andy@petdance.com (Andy Lester) Message-ID: <20060415042924.GA1786@petdance.com> Date: Fri, 14 Apr 2006 23:29:24 -0500 [ 27860] Subject: [PATCH] pp_ctl.c caching From: andy@petdance.com (Andy Lester) Message-ID: <20060417044035.GA28818@petdance.com> Date: Sun, 16 Apr 2006 23:40:35 -0500 Branch: maint-5.8/perl !> dump.c embed.fnc embed.h hv.c mg.c op.c pp_ctl.c proto.h !> thread.h ____________________________________________________________________________ [ 29946] By: nicholas on 2007/01/24 13:23:35 Log: Integrate: [ 27744] Subject: [PATCH] sv.c, rs.t, perlvar.pod (Coverity finding: did you know what happens with $/=\0?) From: jhi@cc.hut.fi (Jarkko Hietaniemi) Message-Id: <20060408152533.C4D5F6D08C@ugli.hut.fi> Date: Sat, 8 Apr 2006 18:25:33 +0300 (EEST) (although I should add that this version of Coverity is actually raising a false positive here, albeit something still interesting) [ 27788] Subject: [PATCH] regexec.c: (Coverity) move NULL check even earlier From: Jarkko Hietaniemi Message-Id: <200604131239.k3DCdXhe124107@kosh.hut.fi> Date: Thu, 13 Apr 2006 15:39:33 +0300 (EEST) [ 27789] Subject: [PATCH] regexec.c: (Coverity) add NULL check From: Jarkko Hietaniemi Message-Id: <200604131252.k3DCqFUL497781@kosh.hut.fi> Date: Thu, 13 Apr 2006 15:52:15 +0300 (EEST) [ 27791] Add another NULL check thanks to Coverity. Branch: maint-5.8/perl !> pod/perlvar.pod regexec.c sv.c t/base/rs.t ____________________________________________________________________________ [ 29945] By: nicholas on 2007/01/24 13:04:13 Log: Integrate: [ 27540] Subject: [PATCH] Data::Dumper consting From: Andy Lester Date: Sun, 19 Mar 2006 01:22:04 -0600 Message-ID: <20060319072204.GB5596@petdance.com> Branch: maint-5.8/perl !> ext/Data/Dumper/Dumper.xs ____________________________________________________________________________ [ 29944] By: nicholas on 2007/01/24 12:25:33 Log: Integrate: [ 27738] ss_dup of SAVEt_COP_ARYBASE can use the same code as SAVEt_I8. [ 27741] As av_dup, gv_dup and hv_dup are the same as sv_dup, code in various branches of Perl_ss_dup() is actually duplicated, so can be merged. [ 27752] Fix pointer error in change 27741, spotted by John E. Malmberg. [ 29082] Merge cases in Perl_ss_dup(). Branch: maint-5.8/perl !> sv.c ____________________________________________________________________________ [ 29943] By: nicholas on 2007/01/24 12:06:54 Log: Integrate: [ 27567] Subject: [PATCH] Re: [perl #38779] NAN's on Win32 From: Dominic Dunlop Date: Wed, 22 Mar 2006 15:18:33 +0100 Message-Id: [ 27576] Subject: RE: [PATCH, no, really!] Re: [perl #38779] NAN's on Win32 From: "Jan Dubois" Date: Wed, 22 Mar 2006 22:49:11 -0800 Message-ID: <060201c64e45$e4ca7020$2217a8c0@candy> Also, back out change #27567. [ 27577] Further adjustement to change #27576 by Jan Dubois (this change being blead-specific, while #27576 is integrable to maint) [ 27637] Fix bug #38815 (localising keys which are UTF-8 encoded didn't delete them correctly on scope exit) [ 27638] Localising hash slices with UTF-8 encoded keys was also buggy. (See also change 27637) [ 27698] Subject: revisited: pow.t test failures on AIX 5.1 perl588 with -Duselongdouble From: "John L. Allen" Date: Thu, 30 Mar 2006 17:53:13 -0500 (EST) Message-ID: [ 27725] Assigning undef to an entry in %ENV shouldn't produce warnings, even though it's silently converted to "" [ 27731] Subject: Re: Smoke [5.9.4] 27728 FAIL(F) MSWin32 WinXP/.Net SP2 (x86/2 cpu) From: Rafael Garcia-Suarez Date: Thu, 6 Apr 2006 14:32:50 +0200 Message-ID: <20060406143250.5a7de4b8@grubert.mandriva.com> [ 27736] Make perl_fini() run when compiling perl with the Sun WorkShop compiler. [ 27756] Subject: [PATCH] pp_sys.c: some Coverity findings: NULL guards for format cv From: jhi@cc.hut.fi (Jarkko Hietaniemi) Message-Id: <20060408155541.46B796D08C@ugli.hut.fi> Date: Sat, 8 Apr 2006 18:55:41 +0300 (EEST) [ 27761] Perl_unsharepvn() was no longer being used in core, and changes to S_unshare_hek_or_pvn() had broken it. Now fixed and tested. Branch: maint-5.8/perl !> README.win32 ext/XS/APItest/APItest.xs ext/XS/APItest/t/hash.t !> hv.c mg.c perl.c pp.c pp_hot.c pp_sys.c scope.c sv.c !> t/lib/warnings/mg t/op/local.t t/op/write.t win32/win32.h ____________________________________________________________________________ [ 29939] By: nicholas on 2007/01/23 23:38:58 Log: Integrate: [ 26835] Additional tests for RT #38207: "Useless localization of constant ($[) in getopts.pl". [ 27495] Subject: [perl #38710] localised stash slice From: Hugo van der Sanden (via RT) Date: Sat, 11 Mar 2006 02:47:56 -0800 Message-ID: (new TODO tests) [ 27547] Subject: [PATCH blead] Re: [perl #38710] localised stash slice From: Rick Delaney Date: Sat, 18 Mar 2006 19:52:11 -0500 Message-ID: <20060319005211.GM25733@localhost.localdomain> One can now localize slices. [ 29938] Add lib to @INC so that ./perl t/op/local.t works. Branch: maint-5.8/perl !> pp.c t/op/local.t ____________________________________________________________________________ [ 29929] By: nicholas on 2007/01/22 23:29:42 Log: Integrate: [ 27671] Abstract all the accesses to cop_arybase (apart from ByteLoader) [ 27674] Convert ByteLoader to use CopARYBASE_set(). [ 27677] Abstract all reads/writes of the hints in COPs with 2 new macros, CopHINTS_get() and CopHINTS_set(). [ 27687] Change 27677 missed two direct accesses to op_private in COPs. I believe that all are now found, as redefining CopHINTS_get(c) to (~(c)->op_private) (with corresponding changes to CopHINTS_set() and the initialisation of PL_compiling) works. Branch: maint-5.8/perl !> bytecode.pl cop.h ext/B/B.xs ext/ByteLoader/bytecode.h !> ext/ByteLoader/byterun.c mg.c op.c perl.h pp.c pp_ctl.c !> pp_hot.c scope.c scope.h toke.c utf8.c utf8.h ____________________________________________________________________________ [ 29928] By: nicholas on 2007/01/22 23:12:43 Log: Integrate: [ 27626] Subject: re-[PATCH] Re: [PATCH] Poison now in two different flavours! From: Jarkko Hietaniemi Date: Sun, 26 Mar 2006 14:53:56 +0300 Message-ID: <442680D4.3000809@gmail.com> [ 27774] Newxz() can use calloc() rather than malloc(), as the implementation of calloc() may know that it has fresh pages from the OS, and so doesn't need to zero them itself. Plus our object code should be slightly smaller. [ 28265] Document PoisonFree() instead of documenting Poison() twice. plus revert change 26776 now that PERL_MEM_LOG is merged: [ 26776] Steve Hay notes that there's a reference PERL_MEM_LOG here, which is inconsistent as PERL_MEM_LOG isn't (yet) merged to maint. Branch: maint-5.8/perl ! pod/perldebguts.pod !> handy.h pod/perlapi.pod pod/perlclib.pod pod/perlhack.pod !> pp_ctl.c scope.c util.c ____________________________________________________________________________ [ 29927] By: nicholas on 2007/01/22 23:11:29 Log: Integrate: [ 27709] Subject: [PATCH] Move conditional SOCKS5 code into #ifdefs From: andy@petdance.com (Andy Lester) Date: Mon, 3 Apr 2006 08:28:18 -0500 Message-ID: <20060403132818.GA7820@petdance.com> [ 27722] Subject: perlio.c breaks on Solaris/gcc when > 256 FDs are available From: Alan Burlison Date: Wed, 05 Apr 2006 15:52:14 +0100 Message-ID: <4433D99E.40800@sun.com> [ 27745] Subject: [PATCH] perlio.c: Coverity finding: finding false (Coverity not able to follow vtable logic) but adding an assert From: jhi@cc.hut.fi (Jarkko Hietaniemi) Message-Id: <20060408161136.CEAC86D08C@ugli.hut.fi> Date: Sat, 8 Apr 2006 19:11:36 +0300 (EEST) [ 27751] Subject: Re: [PATCH] perlio.c: Coverity finding: finding false (Coverity notable to follow vtable logic) but adding an assert From: Jarkko Hietaniemi Date: Sun, 09 Apr 2006 10:18:52 +0300 Message-ID: <4438B55C.9010604@gmail.com> [ 27798] Subject: [PATCH] perlio.c: (Coverity) eliminate temp ptr that confuses Coverity into thinking there is a resource leak From: jhi@cc.hut.fi (Jarkko Hietaniemi) Message-Id: <20060414085644.94A076D08C@ugli.hut.fi> Date: Fri, 14 Apr 2006 11:56:44 +0300 (EEST) [ 27809] Subject: [PATCH] perlio.c: layer data might be allocated and unused (Coverity) From: jhi@cc.hut.fi (Jarkko Hietaniemi) Message-Id: <20060415082417.24F0A6D08C@ugli.hut.fi> Date: Sat, 15 Apr 2006 11:24:17 +0300 (EEST) (with a correction) Branch: maint-5.8/perl !> perlio.c ____________________________________________________________________________ [ 29926] By: nicholas on 2007/01/22 22:47:32 Log: Integrate: [ 27377] Provide Perl_gv_name_set to replace LVALUE use of GvNAME and GvNAMESET. [ 27382] Perl_gv_name_set should not leak the old HEK. Allow the flag GV_ADD to simplify GV initialisation. Branch: maint-5.8/perl ! proto.h !> embed.fnc embed.h global.sym gv.c gv.h sv.c ____________________________________________________________________________ [ 29925] By: nicholas on 2007/01/22 22:10:59 Log: Integrate: [ 27630] Subject: [PATCH] Clean up regexec.c, and remove two functions From: andy@petdance.com (Andy Lester) Date: Wed, 29 Mar 2006 10:13:36 -0600 Message-ID: <20060329161336.GA12456@petdance.com> [ 27641] Subject: [PATCH] Random accumulated patches From: andy@petdance.com (Andy Lester) Date: Thu, 30 Mar 2006 23:42:28 -0600 Message-ID: <20060331054228.GA18940@petdance.com> [ 27649] Subject: [PATCH] Redo PERL_UNUSED_DECL From: andy@petdance.com (Andy Lester) Date: Fri, 31 Mar 2006 00:54:58 -0600 Message-ID: <20060331065458.GA28571@petdance.com> [ 27653] Updated version of the grumpy comment deleted by change #27649, by Jarkko [ 27657] Inconsistencies in paramter const-ness noticed by SADAHIRO Tomoyuki. [ 27690] Subject: [PATCH] Removed unnecessary pointers checks From: andy@petdance.com (Andy Lester) Date: Sat, 1 Apr 2006 01:01:32 -0600 Message-ID: <20060401070132.GA6616@petdance.com> [ 27735] Patch from Jarkko to fix breakage caused by change #27649 see: http://www.xray.mpe.mpg.de/mailing-lists/perl5-porters/2006-04/msg00144.html Branch: maint-5.8/perl ! pp_hot.c pp_sort.c !> cop.h doio.c dump.c embed.fnc embed.h gv.c gv.h hv.c malloc.c !> mg.c op.c perl.c perl.h pp.c pp_ctl.c proto.h regcomp.c !> regexec.c sv.c sv.h toke.c utf8.c ____________________________________________________________________________ [ 29924] By: nicholas on 2007/01/22 20:22:04 Log: Integrate: [ 27617] Merge the cut & paste code from Perl_call_sv/Perl_fold_constants with the near identical code in pp_entertry into Perl_create_eval_scope. Move the cut & paste code from Perl_call_sv/Perl_fold_constants into Perl_delete_eval_scope. [ 29921] Remove duplicate assignment to PL_eval_root in Perl_create_eval_scope Branch: maint-5.8/perl ! perl.c pp_ctl.c !> cop.h embed.fnc embed.h ext/B/t/concise-xs.t op.c proto.h ____________________________________________________________________________ [ 29920] By: nicholas on 2007/01/22 19:20:43 Log: Integrate: [ 27668] If the passed in target rv for Perl_newSVrv() was already a reference to something, it would leak the reference count on that thing. [ 27676] 'SvIOK(sv) ? SvIVX(sv) : sv_2iv(sv)' is 'SvIV(sv)' so write it as such. [ 27696] Take splint into account for PERL_UNUSED_ARG (patch by Jarkko) [ 27723] Initializing variables is good. This is another Coverity catch. [ 27724] Try not to use negative values when accessing arrays in C. Yet another Coverity catch. [ 27726] Perl_save_alloc can use the new(er) SSGROW rather than looping. [ 27750] Subject: Re: [PATCH] locale.c: Coverity finding From: Jarkko Hietaniemi Date: Sun, 09 Apr 2006 10:31:32 +0300 Message-ID: <4438B854.6040301@gmail.com> [ 27755] Subject: [PATCH] pp_sys.c: some Coverity findings: NULL guards for io pointers From: jhi@cc.hut.fi (Jarkko Hietaniemi) Message-Id: <20060408155232.4C9726D08C@ugli.hut.fi> Date: Sat, 8 Apr 2006 18:52:32 +0300 (EEST) [ 27757] Subject: [PATCH] Fix coverity warning in utf8.c From: Andy Dougherty Date: Mon, 10 Apr 2006 11:18:35 -0400 (EDT) Message-ID: Branch: maint-5.8/perl !> XSUB.h locale.c mg.c perl.h pp_sys.c scope.c sv.c utf8.c !> util.c ____________________________________________________________________________ [ 29919] By: nicholas on 2007/01/22 18:47:36 Log: Integrate: [ 27608] Add S_space_join_names_mortal() which joins a char** array with " "s, replacing 5 instances of the same code. [ 27612] Replace all sv_mortalcopy(&PL_sv_no); sv_set*(...) with sv2mortal(newSV*(...)) to avoid needless upgrades to PVNV and needless allocation of a copy of "". Branch: maint-5.8/perl !> embed.fnc embed.h pp_sys.c proto.h ____________________________________________________________________________ [ 29918] By: nicholas on 2007/01/22 17:29:45 Log: Update Changes. Branch: maint-5.8/perl ! Changes patchlevel.h ____________________________________________________________________________ [ 29916] By: nicholas on 2007/01/22 16:26:58 Log: Integrate: [ 27526] reduce S_regrepeat_hard() callers from 3 to 1 [ 27534] inline, then delete, S_regrepeat_hard() [ 27535] Restore a bit of change 27533 that change 27534 inadvertently unwound. [ 27569] remove idential code branch from regmatch() Branch: maint-5.8/perl !> embed.fnc embed.h proto.h regexec.c ____________________________________________________________________________ [ 29915] By: nicholas on 2007/01/22 15:49:59 Log: Integrate: [ 27533] sv_find() returning false, followed by sv_magic() to add the magic, followed immediately by sv_find() to find it, is somewhat wasteful. So use sv_magicext(). (All cases are also correct w.r.t. SvREADONLY()) [ 27542] Subject: [PATCH] Change the semantics of S_isa_lookup From: Andy Lester Date: Sat, 18 Mar 2006 00:28:45 -0600 Message-ID: <20060318062845.GA11607@petdance.com> [ 27548] If the scalar has just been upgraded to SVt_RV, there's no way SvPVX can be non-NULL, so no need to check. Branch: maint-5.8/perl !> embed.fnc mg.c pp.c pp_ctl.c pp_hot.c proto.h regexec.c sv.c !> universal.c ____________________________________________________________________________ [ 29914] By: nicholas on 2007/01/22 15:29:10 Log: Integrate: [ 27439] MAD changes for bare skipspace() [ 27442] Replace PL_nextval[PL_nexttoke].opval with the macro NEXTVAL_NEXTTOKE as MAD conditionally changes this consistently in 27 places. Branch: maint-5.8/perl ! toke.c ____________________________________________________________________________ [ 29913] By: nicholas on 2007/01/22 14:45:23 Log: Integrate: [ 27213] The two whitespace changes from Subject: arena-rework : consolidated patch From: Jim Cromie Message-ID: <43F0F649.9040205@gmail.com> Date: Mon, 13 Feb 2006 14:12:41 -0700 [ 27397] Subject: Re: [PATCH] Removed a redundant o->op_type From: Paul Johnson Date: Tue, 28 Feb 2006 20:17:31 +0100 Message-ID: <20060228191731.GD20933@pjcj.net> [ 27414] Subject: [PATCH] Removed unused check and assignment in pp_hot.c From: andy@petdance.com (Andy Lester) Date: Wed, 8 Mar 2006 00:45:29 -0600 Message-ID: <20060308064529.GA16499@petdance.com> [ 27423] Change from Larry in the MAD code which seems to be unrelated to MAD. [ 27443] Refactoring from Larry implicit in the MAD patch. [ 27444] Another refactoring from Larry implicit in the MAD patch. [ 27451] Avoid scribbling on the passed in command line string for -i in moreswitches. [ 27452] Superior *printf-fu lets us avoid modifying source code. [ 27456] Correct another gv_fetchpv() - it takes a bitmask of flags, not simply TRUE/FALSE. [ 27458] Tease apart the keyword/subroutine test in S_checkcomma. [ 27459] Summon constman! S_checkcomma now has all 3 arguments const char. [ 27469] Subject: [PATCH] Localizing some vars in S_checkcomma From: andy@petdance.com (Andy Lester) Message-ID: <20060310223100.GA13455@petdance.com> Date: Fri, 10 Mar 2006 16:31:00 -0600 [ 27470] Subject: [PATCH] Sync up embed.fnc and doio.c From: andy@petdance.com (Andy Lester) Message-ID: <20060311071338.GA12490@petdance.com> Date: Sat, 11 Mar 2006 01:13:38 -0600 [ 27480] isSPACE('\0') is false, so can simplify the condition in the for loop. As the for loop is basically a while loop, write it as one, to make its intentions clearer. [ 27493] Subject: [PATCH] Do nothing in a GCC-friendly way From: andy@petdance.com (Andy Lester) Date: Mon, 13 Mar 2006 12:48:50 -0600 Message-ID: <20060313184850.GA16905@petdance.com> [ 27507] Expunge references to HV_AMAGIC, HV_AMAGICbad and HV_AMAGICmb which have been unused since revision 1. [ 27523] Correct the macros in the comments in hv.c. Given the improvements in compiler optimisation, and the almost irrelevance of instructions executed over cache misses, maybe we should just remove the comments by putting their code back in place of the o{ptimis,bfuscat}ed version. Branch: maint-5.8/perl ! proto.h !> embed.fnc gv.c hv.c hv.h op.c perl.c pp_hot.c sv.h toke.c ____________________________________________________________________________ [ 29912] By: nicholas on 2007/01/21 23:33:27 Log: Integrate: [ 26924] Un-const magic virtual tables; this consting breaks the nasty things done by Swig, as noticed by Audrey. [ 27361] Subject: [PATCH] mgvtbl initialization From: andy@petdance.com (Andy Lester) Date: Wed, 1 Mar 2006 22:52:20 -0600 Message-ID: <20060302045220.GA15985@petdance.com> [ 28599] Subject: [PATCH] make magic vtables const if PERL_GLOBAL_STRUCT_PRIVATE From: Jarkko Hietaniemi Date: Mon, 17 Jul 2006 09:09:24 +0300 Message-ID: <44BB2994.5090609@iki.fi> [ 29910] Restore the const to vtbl in Perl_mg_localize() removed in change 26924. [ 29911] Assignments to function pointers can't be NULL, as that's a pointer to data, and assigning pointers to data to pointers to functions is not ANSI-conformant. Branch: maint-5.8/perl ! mg.c !> embed.fnc mg.h perl.h pod/perlapi.pod pod/perlguts.pod proto.h !> sv.c ____________________________________________________________________________ [ 29909] By: nicholas on 2007/01/21 22:28:56 Log: Integrate: [ 27327] Abstract the specific use of SvCUR in GVs for detecting variables on both sides of an assignment using GvASSIGN_GENERATION() and GvASSIGN_GENERATION_set(). [ 27337] Fix code-before-declaration error (under VC++) from change #27334 [ 27349] Subject: [PATCH] Removed a redundant o->op_type From: andy@petdance.com (Andy Lester) Message-ID: <20060227202043.GA7783@petdance.com> Date: Mon, 27 Feb 2006 14:20:43 -0600 [ 27350] In Perl_sv_setsv_flags, swap the default in the type based switch to the "Bizarre copy" failure case. [ 27351] Silence a VC++ warning with DEBUGGING builds (signed/unsigned mismatch) [ 27355] Subject: [PATCH] a buglet in fakesdio.h From: Anton Berezin Date: Wed, 1 Mar 2006 16:56:13 +0100 Message-ID: <20060301155613.GA30368@heechee.tobez.org> [ 27360] Subject: [PATCH] Non-null optimizations From: andy@petdance.com (Andy Lester) Date: Wed, 1 Mar 2006 22:13:45 -0600 Message-ID: <20060302041345.GA13731@petdance.com> Branch: maint-5.8/perl !> embed.fnc fakesdio.h gv.h op.c perlio.c pod/perlapi.pod !> proto.h sv.c ____________________________________________________________________________ [ 29908] By: nicholas on 2007/01/21 21:58:40 Log: Integrate: [ 27270] Turn on match string copying when /e flag is set on a substitution. Subject: [PATCH] dodge a valgrind error (for maint or blead) From: Jarkko Hietaniemi Date: Feb 22, 2006 8:16 PM Message-ID: <43FCB896.7060106@gmail.com> [ 27320] Given that Perl_gp_free() is refcount-dec-and-maybe-free for the glob pointer, it's inconsistent that it only sets this GV's GvGP to 0 if this GV happened to have the last reference. Why should this GV care if it had the last reference? So always set it to 0. It's free. Gone. [ 27359] Don't put strings with embedded NULs in the environment. This makes things like -d:Foo=bar work again. [ 27402] $[ = 2 should not warn. [ 27403] This looks like a memory leak. [ 27436] Fix a memory leak in ck_grep(), spotted by coverity: perl -e'eval "grep" while 1' [ 27448] The Coverity audit is upset that the scream_olds variable is not directly initialized, although it is likely initialized after another variable takes a reference to it. [ 27476] Possible NULL pointer reference found by Coverity checks. [ 27477] Perl_newWHILEOP() had exactly the same lack of a NULL pointer check that Perl_newLOOPOP() had and fixed in change #27476. Maybe some refactoring is needed? [ 27515] require should ignore directories found when searching @INC not just die as soon as it finds one. It should for instance be possible to for require "File" to read the file "./File" even if there happens to be a "File" directory in perl's standard library. This fixes the RT #24404 fix in change 26373. Branch: maint-5.8/perl !> gv.c op.c perl.c pp_ctl.c pp_hot.c regexec.c t/comp/require.t !> t/lib/warnings/op ____________________________________________________________________________ [ 29906] By: nicholas on 2007/01/21 12:16:40 Log: Integrate: [ 27306] Replace usage of GvSTASH for storing the stash of C with OURSTASH. Set the stash with OURSTASH_SET. [ 29633] With PAD_COMPNAME_GEN in SvUVX, SvCUR is trustworthy once more. [just the pad.h changes] [ 29679] Rename OURSTASH to SvOURSTASH and OURSTASH_set to SvOURSTASH_set. Branch: maint-5.8/perl ! pad.c !> dump.c pad.h sv.c sv.h ____________________________________________________________________________ [ 29904] By: nicholas on 2007/01/21 00:30:14 Log: Integrate: [ 27301] Subject: [PATCH] Perl_save_aelem can't be const From: Andy Lester Date: Thu, 23 Feb 2006 22:53:40 -0600 Message-ID: <20060224045340.GA6318@petdance.com> [ 27339] ptr_table_find() is only defined with ithreads Branch: maint-5.8/perl !> embed.fnc embed.h proto.h scope.c sv.c ____________________________________________________________________________ [ 29903] By: nicholas on 2007/01/21 00:16:12 Log: Integrate: [ 27294] Replace direct flags tests & manipulations for SVpad_TYPED and SVpad_OUR with macros SvPAD_TYPED(), SvPAD_OUR() etc, to abstract away the flags bits acutally used to store this information. Branch: maint-5.8/perl !> XSUB.h dump.c op.c pad.c pad.h pp.c sv.c sv.h toke.c ____________________________________________________________________________ [ 29902] By: nicholas on 2007/01/20 23:57:28 Log: Integrate: [ 27271] Add a test for study() on tied scalars, by Andy Lester after Rick Delaney [ 27272] Mark some new study tests as TODO [ 27273] Avoid Cing any strings that might change underneath us, such as tied scalars and scalars with overloaded stringification. [ 27274] Fix typos and a missing bracket. Branch: maint-5.8/perl +> t/op/studytied.t !> MANIFEST pp.c ____________________________________________________________________________ [ 29901] By: nicholas on 2007/01/20 23:44:56 Log: Integrate: [ 27265] Can merge the two arms of Perl_magic_getglob to save space. [ 27268] Test warnings for converting globs to other forms. [ 27269] sv_dump should report the PV for PVGVs, as it can get set. (Typeglobs stringify via a call to GET magic, which uses the PV slot.) [ 27322] With PERL_DONT_CREATE_GVSV, don't need to create a new GvSV in Perl_save_scalar. Branch: maint-5.8/perl ! dump.c !> ext/Devel/Peek/t/Peek.t mg.c scope.c t/op/gv.t ____________________________________________________________________________ [ 29899] By: nicholas on 2007/01/20 23:15:32 Log: Integrate: [ 27258] Document the many uses of SVf_FAKE. [ 27259] Re-order the definitions of the private bits in SvFLAGS by value, so that we can see where we're already double-booked. [ 27293] Document a sixth use for SVf_FAKE [ 29554] Turn a tab into 2 spaces to make more room for comments about SV flags. Sorry about damaging the change history - dig through this change with http://public.activestate.com/cgi-bin/perlbrowse?show_blame=Show+Annotated+File&filename=sv.h&rev=&change=29552 [ 29559] Better documentation for SVf_UTF8. Including SvPV() 1st, SvUTF() 2nd. Branch: maint-5.8/perl !> sv.h ____________________________________________________________________________ [ 29898] By: nicholas on 2007/01/20 18:43:49 Log: Integrate: [ 27241] Add a new CvISXSUB() macro, for abstracting the test as to whether a PVCV is perl or XS. [ 27267] Change 24643 made the mistake of assuming that CvCONST can only be true on XSUBs. Somehow it can also end up on perl subs. Bug spotted by and test case from Marcus Holland-Moritz. Branch: maint-5.8/perl !> cv.h dump.c ext/Devel/Peek/Peek.xs gv.c op.c pad.c pp_ctl.c !> pp_hot.c pp_sort.c sv.c t/op/threads.t ____________________________________________________________________________ [ 29897] By: nicholas on 2007/01/20 18:14:46 Log: Integrate: [ 27178] Remove a my_perl that slipped in [ 27300] Subject: [PATCH] unused context warnings From: andy@petdance.com (Andy Lester) Date: Tue, 21 Feb 2006 00:27:11 -0600 Message-ID: <20060221062711.GA16160@petdance.com> Branch: maint-5.8/perl !> deb.c embed.pl hv.c locale.c mathoms.c mg.c miniperlmain.c !> numeric.c op.c perl.c perl.h perlapi.c perlio.c pp_ctl.c !> regcomp.c scope.c sv.c toke.c universal.c utf8.c util.c ____________________________________________________________________________ [ 29896] By: nicholas on 2007/01/20 17:47:00 Log: Integrate: [ 27177] Subject: [PATCH] Sun Studio lint patches From: andy@petdance.com (Andy Lester) Date: Mon, 13 Feb 2006 23:46:17 -0600 Message-ID: <20060214054617.GA8824@petdance.com> Branch: maint-5.8/perl !> Makefile.SH gv.c hv.c op.c pad.c perlio.c pp.c pp_ctl.c !> pp_hot.c regcomp.c sv.c toke.c utf8.c ____________________________________________________________________________ [ 29891] By: nicholas on 2007/01/20 00:30:02 Log: Integrate: [ 27215] Subject: arena-rework : consolidated patch From: Jim Cromie Message-ID: <43F0F649.9040205@gmail.com> Date: Mon, 13 Feb 2006 14:12:41 -0700 Tweaked somewhat to split the arena boolean from the arena_size, and with the PTE still doubling-up with one of the SV types in the array. [ 27290] Using U8 rather than size type shrinks the body_details table. [ 27291] Shrink struct body_details and hence sv.o slightly further by using bitfields. [ 27292] We have sufficient spare bits to store the SV type in body_details, so add a small sanity check to ensure that we have the array in the correct order. [ 27338] Workaround for initialization errors on HP's pre-compiler Do not ask me to explain. This re-definition works on AIX and HP-UX. Let's await the smokes. [ 27405] Remove the #define - ARENASETS are always on now. [ 29878] Move C inside the only function that uses it. [ 29879] As Perl_get_arena() is dealing with sizes, use size_t rather than int, as it's both unsigned and semantically the correct width for a size. As all arenas get cleared later on, can use Newx() rather than Newxz(). [ 29881] I think that it's clearer if aroot is only struct arena_set * (rather than ** and constantly dereferneced). Move the declaration of new_root into the block it is used in. Add a comment describing where the arenas for arena_sets themselves aren't rooted anywhere. [ 29882] Add a parameter to Perl_get_arena() to pass in the SV type, and record this in the arena description. Change all sizes to unsigned values. Make Perl_sv_free_arenas() loop downwards to free memory, simplifying the logic. Remove my erroneous comment added in change 29881. Branch: maint-5.8/perl ! sv.c !> embed.fnc embed.h hv.c perl.h pod/perlapi.pod proto.h sv.h ____________________________________________________________________________ [ 29890] By: nicholas on 2007/01/20 00:10:08 Log: Integrate: [ 27172] Teach checkAUTHORS.pl about Changelogs with leading tabs in the log entries, and fix a malformed log entry in Changes5.005 Branch: maint-5.8/perl !> Changes5.005 Porting/checkAUTHORS.pl ____________________________________________________________________________ [ 29889] By: nicholas on 2007/01/19 23:58:44 Log: Integrate: [ 27399] FAQ sync. (FAQ now moved to svn) [ 27905] Changes to perlfaq8 "How do I find out if I'm running interactively or not?" suggested by Slaven Rezic in RT #38901: perlfaq8 correction [ 28464] s/have a more/have more/ typo fix in perlfaq4 [ 28820] FAQ sync [ 29034] FAQ sync [ 29282] FAQ sync Branch: maint-5.8/perl !> pod/perlfaq.pod pod/perlfaq1.pod pod/perlfaq2.pod !> pod/perlfaq3.pod pod/perlfaq4.pod pod/perlfaq5.pod !> pod/perlfaq6.pod pod/perlfaq7.pod pod/perlfaq8.pod !> pod/perlfaq9.pod ____________________________________________________________________________ [ 29888] By: nicholas on 2007/01/19 21:24:46 Log: Integrate: [ 26545] Marcus Holland-Moritz suggested that assert should Perl_croak_nocontext rather than Perl_croak. [ 27136] Subject: [PATCH] Trying my "remove the pTHXes" patch again From: Andy Lester Date: Thu, 9 Feb 2006 09:40:18 -0600 Message-ID: <20060209154018.GA14610@petdance.com> [ 27137] Regenerated headers for change #27136. [ 27138] Really regenerate headers [ 27140] Fix breakage in S_set_csh() from change #27136. [ 27145] Change 27136 seems to have inadvertently upset some compilers, presumably where NULL was defined as 0, not (void *)0 [ 27148] Fix the effects of 27136 on Win32 with DEBUGGING and PERL_IMPLICIT_SYS [ 27152] A better fix than 27148 Subject: Win32 assert (was Re: Smoke [5.9.4] 27145 FAIL(M) MSWin32 WinXP/.Net SP2 (x86/2 cpu)) From: Nicholas Clark Date: Fri, 10 Feb 2006 14:58:47 +0000 Message-ID: <20060210145847.GA85020@plum.flirble.org> Subject: RE: Win32 assert (was Re: Smoke [5.9.4] 27145 FAIL(M) MSWin32 WinXP/.Net SP2 (x86/2 cpu)) From: "Jan Dubois" Date: Fri, 10 Feb 2006 09:03:41 -0800 Message-ID: <024401c62e63$f1e53c10$6062a8c0@candy> Branch: maint-5.8/perl ! embed.fnc embed.h proto.h sv.c !> Makefile.SH cc_runtime.h cop.h deb.c hv.c mg.c op.c op.h !> perl.c perl.h pp.c pp_ctl.c pp_sort.c regcomp.c regexec.c !> toke.c utf8.c util.c ____________________________________________________________________________ [ 29886] By: nicholas on 2007/01/19 20:40:06 Log: Integrate: [ 27079] Subject: [patch] arena rework - arena sets From: Jim Cromie Message-ID: <43DF4F66.4080808@gmail.com> Date: Tue, 31 Jan 2006 04:52:06 -0700 [ 27080] Tweak arena sets to avoid assignment between different pointers. [ 27081] Rejig the definition of ARENAS_PER_SET to maximise the size of the array whilst keeping the structure within PERL_ARENA_SIZE. [ 27097] Subject: [patch] rework arenas - repair arenasets From: Jim Cromie Message-ID: <43E5F012.1080203@gmail.com> Date: Sun, 05 Feb 2006 05:31:14 -0700 Branch: maint-5.8/perl !> embed.fnc embed.h hv.c proto.h sv.c ____________________________________________________________________________ [ 29874] By: nicholas on 2007/01/18 16:26:05 Log: Integrate: [ 27699] Subject: [PATCH] ext/IO/t/io_unix.t From: Jarkko Hietaniemi Date: Sun, 02 Apr 2006 21:57:19 +0300 Message-ID: <44301E8F.2060503@gmail.com> [ 29578] Subject: [PATCH 5.8.8] Fixes for the test suite on OS/2 From: Ilya Zakharevich Date: Wed, 13 Dec 2006 18:32:22 -0800 Message-ID: <20061214023222.GA29084@powdermilk.math.berkeley.edu> Branch: maint-5.8/perl !> ext/IO/t/io_unix.t lib/Time/Local.t os2/OS2/ExtAttr/t/os2_ea.t !> t/io/fs.t t/op/stat.t ____________________________________________________________________________ [ 29873] By: nicholas on 2007/01/18 16:11:31 Log: Integrate: [ 28187] Subject: t/uni/overload.t From: Jarkko Hietaniemi Message-ID: <4465988C.9080407@gmail.com> Date: Sat, 13 May 2006 11:27:56 +0300 [ 28575] overload.t has same locale problem with VMS that it has with dec_osf [ 29099] Silence an expected warning (when playing with utf8ness and string overloading) Branch: maint-5.8/perl !> t/uni/overload.t ____________________________________________________________________________ [ 29870] By: nicholas on 2007/01/18 12:45:46 Log: Integrate: [ 24755] s/PERL_COPY_ON_WRITE/PERL_OLD_COPY_ON_WRITE/g Branch: maint-5.8/perl !> ext/B/t/bytecode.t ____________________________________________________________________________ [ 29869] By: nicholas on 2007/01/18 12:25:16 Log: Integrate: [ 26866] Upgrade to Devel::PPPort 3.07 [ 26913] find_perl() must be after environment initialization [ 26914] Upgrade to Devel::PPPort 3.08 [ 28259] Upgrade to Devel::PPPort 3.08_01 [ 28267] Upgrade to Devel::PPPort 3.08_02 [ 28307] Upgrade to Devel::PPPort 3.08_03 [ 28332] Upgrade to Devel::PPPort 3.08_04 [ 28420] Upgrade to Devel::PPPort 3.08_05 [ 28424] Upgrade to Devel::PPPort 3.08_06 [ 28474] Upgrade to Devel::PPPort 3.08_07 [ 28507] Upgrade to Devel::PPPort 3.09 [ 28605] Upgrade to Devel::PPPort 3.09_01 [ 28616] Upgrade to Devel::PPPort 3.09_02 [ 28714] Upgrade to Devel::PPPort 3.10 Branch: maint-5.8/perl +> (branch 28 files) - ext/Devel/PPPort/MANIFEST ext/Devel/PPPort/PPPort.pm !> (integrate 89 files) ____________________________________________________________________________ [ 29865] By: nicholas on 2007/01/17 23:48:41 Log: Integrate: [ 26670] Make '-s' on the shebang line able to parse -foo=bar switches again. This feature was broken by change 19695 some years ago and integrated into perl-5.8.1. perl-5.8.0 was fine. Ref http://bugs.activestate.com/show_bug.cgi?id=43483 Branch: maint-5.8/perl !> perl.c t/run/switches.t ____________________________________________________________________________ [ 29864] By: nicholas on 2007/01/17 23:29:13 Log: Integrate: [ 24912] Same fix, for the test, now. [ 26321] Disallow sprintf's vector handling for non-integer formats. Avoids core dump for printf("%vs") and similar. [ 26326] Drop "v" prefix from sprintf("%vd", $^V). The sprintf documentation has this example: printf "version is v%vd\n", $^V; and it printed 'version is vv5.9.3\n'. [ 26327] Oops, change 26326 broke t/run/switches.t Reverting change 24912 fixes it. [ 26339] Add overflow check to EXPECT_NUMBER() used by sv_vcatpvfn(). sprintf() or printf() will now croak if any of the indexes and widths specified in the format string are too large. [ 26342] Remove tests that were not portable to 64bit ints. Branch: maint-5.8/perl ! t/op/sprintf.t !> pod/perldiag.pod sv.c t/op/sprintf2.t t/run/switches.t ____________________________________________________________________________ [ 29863] By: nicholas on 2007/01/17 22:55:14 Log: Two regenerated headers. Branch: maint-5.8/perl ! embedvar.h perlapi.h ____________________________________________________________________________ [ 29862] By: nicholas on 2007/01/17 22:53:30 Log: Integrate: [ 26296] Subject: [PATCH] All args in embed.fnc should be named From: Andy Lester Date: Wed, 7 Dec 2005 10:08:18 -0600 Message-ID: <20051207160818.GA13465@petdance.com> Branch: maint-5.8/perl ! proto.h !> embed.fnc embed.pl pod/perlapi.pod ____________________________________________________________________________ [ 29861] By: nicholas on 2007/01/17 22:36:16 Log: Integrate: [ 26881] The current OP that is tested in doeval() can be a require(), an eval() or a do(). When we test whether OPf_SPECIAL is set on it, be sure to take into account that it's the eval-related meaning of OPf_SPECIAL we're testing for, not the require-related one. [ 26883] Regression test for change #26881 [ 26891] New t/comp/require.t tests should go before the conditional exit. Branch: maint-5.8/perl !> pp_ctl.c t/comp/require.t ____________________________________________________________________________ [ 29860] By: nicholas on 2007/01/17 22:08:43 Log: Remove the new assertion integrated from blead. It might just not be safe. Branch: maint-5.8/perl ! sv.c ____________________________________________________________________________ [ 29859] By: nicholas on 2007/01/17 22:07:40 Log: Integrate: [ 27126] Merging pp_bit_or and pp_bit_xor shrinks the object code by about .7K. The overloading tests are not free. [ 27127] Borland's C compiler warns that the & is unnecessary. [ 27129] Subject: [PATCH] Handle unused args From: andy@petdance.com (Andy Lester) Date: Tue, 7 Feb 2006 16:36:56 -0600 Message-ID: <20060207223656.GA5177@petdance.com> [ 27134] Avoid calling all the scanning code in gv_fetchpvn_flags when saving $1 etc, as we can perfom a hash lookup on %:: directly. [ 27161] Remove SOFT_CAST() as it no longer does anything useful. [ 27194] Get perl_fini() running on HP-UX again. It was effectively disabled by change 24667 since __ux_version is only available under DCE threads which isn't used when perl is built on HP-UX 11 and later. [ 27203] Subject: Problem compiling swigged c++ code with 5.8.8 From: Merijn Broeren Date: Wed, 15 Feb 2006 13:51:49 +0100 Message-ID: <20060215125148.GA12535@brugman.iloquent.nl> [ 27227] SvROK(sv) will never be true when SvIOKp() or SvNOKp() is true, so the code inside the if() test in sv_2pv_flags is dead and can be removed. [ 27235] Remove the last (U16) cast for CV depths, missed by change 17835. [ 27279] Remove un-needed case in Perl_sv_setsv_flags (sv_upgrade will cover this one for us). [ 27282] Recalculate dstr in Perl_sv_setsv_flags, as dstr may have been upgraded. [ 27288] In XS_attributes__guess_stash, attempting to call Gv* on a PVMG is bad and wrong. Branch: maint-5.8/perl !> XSUB.h doio.c mathoms.c opcode.h opcode.pl pad.c perl.c perl.h !> pp.c pp.h pp_ctl.c pp_sys.c regcomp.c scope.h sv.c xsutils.c ____________________________________________________________________________ [ 29858] By: nicholas on 2007/01/17 21:17:52 Log: Integrate: [ 26989] Call gv_fetchpvn_flags where we already know the length. [ 27005] Avoid a warning from the Irix C compiler. [ 27006] Silence another Irix compiler warning. [ 27014] Silence a warning from the MS compiler about signed/unsigned mismatch. [ 27056] Death to magic characters! No pun intended. [ 27064] Fix file where internal XS functions are defined [ 27086] bcc was issuing a warning about this line in Perl_sv_2cv, so tweak it to keep it happy. [ 27089] Without this Windows and Netware compilers should be complaining about perl.c having code before a declaration, *but only if compiling with PERL_TRACK_MEMPOOL*. So, will it fix the smoke failures? Place bets now. [ 27093] Yes, I broke Win32 with change 27089 by having a { in the wrong place. [ 27102] Subject: [PATCH] Cleaning up shadowed variables From: Andy Lester Date: Sun, 5 Feb 2006 21:56:43 -0600 Message-Id: [ 27112] Given that the memory allocated in Perl_bytes_from_utf8 and Perl_bytes_to_utf8 will immediately be written to, I see no need to allocate it zeroed. Branch: maint-5.8/perl ! toke.c !> doio.c ext/B/B.xs hv.c op.c perl.c pp_ctl.c sv.c universal.c !> utf8.c util.c ____________________________________________________________________________ [ 29856] By: nicholas on 2007/01/17 19:49:29 Log: Integrate: [ 26981] The flags manipulation in sv_setsv_flags can be more efficient. [ 26982] Merge the IOK and NOK clauses together in sv_setsv_flags. [ 26987] Squeeze more flag manipulations together in sv_setsv_flags. [ 26997] Assert that IVs and NVs can never be tainted. [ 27043] Correct my comment, so that it's actually useful. :-) [ 27107] Remove a duplicate flag copy line from Perl_sv_setsv_flags. [ 27110] I think that "merge Perl_sv_2[inpu]v" and "reduce duplication in sv_setsv_flags" are about as done as they can be. [ 29855] Replace SvRELEASE_IVX(dstr) with SvOOK_off(dstr) in sv_setsv_flags(), because it's not possible for dstr to be COW at this point, due to an earlier force_normal. Branch: maint-5.8/perl ! sv.c !> pod/perltodo.pod ____________________________________________________________________________ [ 29854] By: nicholas on 2007/01/17 18:52:01 Log: Integrate: [ 27066] It's actually easier to get rid of PL_fdscript than we thought. [ 27068] Oops. I *thought* that I had checked that all changed files were open. Clearly not. (Fixes change 27066) [ 27069] Passing the flag to S_forbit_setid as a single char rather than a string is a slight space optimisation. [ 27070] Eliminating PL_suidscript is more tricky, and requires changing the prototype of Perl_moreswitches. [ 27072] Change 27070 failed to initialise a variable. (But despite that still passed all tests locally.) [ 27073] Pull the variables fdscript and suidscript into a much smaller scope. Replace suidscript with literal -1 where it could only have been -1. (Re-indent the other code inside the new block creating the scope. Rejig the indentation of the conditionals) [ 27074] The MSDOS-only call to moreswitches will never reach forbid_setid, so it can have a parameter of -1 without changing any behaviour. I see no reason to forbid "command line" switches when they are actually coming from the #! line of the script itself (as read by perl) as they must have been written by the owner of the script, rather than being crafted by the user manipulating the command line, and therefore cannot be subverted. (Or alternatively, can be subverted as easily as the rest of the script, and we're permitting *that* to be run) [ 27075] And as we've now got to the point where all calls to Perl_moreswitches have -1 as the second argument, we can remove the second argument, which gets us back to where we started, only with the elimination of a per-thread variable. [ 27076] S_find_beginning is only called from one place, so by hoisting the forbid_setid check for -x into the caller, we can eliminate the suidscript parameter from S_find_beginning. [ 27077] Remove obsolete comment. [ 27087] Fixed embed.fnc entry for S_forbid_setid() and picked up a change in pod/perlapi.pod for good measure when regenerating files. Branch: maint-5.8/perl !> embed.fnc embed.h embedvar.h intrpvar.h perl.c perlapi.h !> pod/perlapi.pod proto.h toke.c ____________________________________________________________________________ [ 29851] By: nicholas on 2007/01/17 15:41:23 Log: Integrate: [ 27104] Fix bug #38439 - reference to typeglob assignment needs to be based on SvROK(sstr) rather than SvTYPE(sstr) == SVt_RV. [ 27111] Rename S_glob_assign to S_glob_assign_glob Rename S_pvgv_assign to S_glob_assign_ref I don't think that they share enough common code to merge them. Branch: maint-5.8/perl !> sv.c t/op/gv.t ____________________________________________________________________________ [ 29850] By: nicholas on 2007/01/17 14:15:59 Log: Integrate: [ 26978] Pull the other large chunk of PVGV assignment code into S_pvgv_assign. There seems to be some duplication with S_glob_assign. [ 26991] C compilers don't let you return void. Pay attention gcc, and stop aping Shildt. [ 27098] Start to merge the 6 arms of S_pvgv_assign into one. [ 27099] Merge the default case (GvSV) with the common code in S_pvgv_assign. [ 27100] Merge the PVAV and PVHV clauses into the common code of S_pvgv_assign. [ 27101] Merge the PVCV case into the common code in S_pvgv_assign. Branch: maint-5.8/perl !> sv.c ____________________________________________________________________________ [ 29849] By: nicholas on 2007/01/17 13:51:31 Log: Integrate: [ 26976] Pull glob_assign out from sv_setsv_flags into a static function. [ 26977] Remove goto glob_assign; Branch: maint-5.8/perl !> sv.c ____________________________________________________________________________ [ 29847] By: nicholas on 2007/01/17 12:30:43 Log: Integrate: [ 27027] Intel C Compiler 9.0 hatefully defines __GNUC__. Unfortunately, icc does not support gcc built-ins like __attribiute__. [ 27568] Fix some test failures in t/op/cmp.t when compiling with the Intel C++ Compiler on Linux. [ 28566] Keep Intel C++ from claiming that it is gcc. [ 28728] Additional floating point strictness is needed to get Intel cc to pass its tests. Branch: maint-5.8/perl !> hints/linux.sh perl.h ____________________________________________________________________________ [ 29846] By: nicholas on 2007/01/17 11:36:40 Log: Integrate: [ 26474] Create a struct to use as the header with PERL_TRACK_MEMPOOL, so that other information can be stored in it. [ 26476] If PERL_TRACK_MEMPOOL and PERL_POISON are in use, then scribble all over memory to invalidate it just before free()ing it. [ 26478] For PERL_TRACK_MEMPOOL with PERL_POISON, Poison the end of any block being shrunk in realloc() [ 27084] Enhance PERL_TRACK_MEMPOOL so that it also emulates the PerlHost behaviour of freeing up all memory at thread exit. With this and tools such as valgrind you will now get warnings as soon as you read from the deallocated memory, rather than just a warning much later about freeing to the wrong pool. [ 27088] Don't export PL_memory_debug_header to Win32 (and others) unless PERL_TRACK_MEMPOOL is defined. [ 27151] With PERL_POISON defined, ensure freshly malloc()ed memory isn't zeros, and when PERL_TRACK_MEMPOOL is also defined scribble on any extension added by realloc(). [ 27343] Subject: [PATCH] PERL_TRACK_MEMPOOL cripples environment after exit() From: Marcus Holland-Moritz Date: Sun, 26 Feb 2006 20:47:21 +0100 Message-ID: <20060226204721.00be2bff@r2d2> [ 27396] Don't free thread memory if PERL_DESTRUCT_LEVEL is set to a non-zero value as we're probably hunting memory leaks then Branch: maint-5.8/perl ! perl.h !> embedvar.h intrpvar.h makedef.pl miniperlmain.c perl.c !> perlapi.h pod/perltodo.pod sv.c util.c ____________________________________________________________________________ [ 29840] By: nicholas on 2007/01/16 08:45:48 Log: Update Changes Branch: maint-5.8/perl ! Changes patchlevel.h ____________________________________________________________________________ [ 29832] By: nicholas on 2007/01/15 16:30:37 Log: Avoid a SEGV in DBI's test suite, discovered by Andreas. Probably need to go through and audit to see if there are analagous places that need tweaking. Branch: maint-5.8/perl ! perl.c ____________________________________________________________________________ [ 29809] By: nicholas on 2007/01/14 13:47:07 Log: Integrate: [ 26983] Add gv_fetchpvs, which uses STR_WITH_LEN to call gv_fetchpvn_flags. Many strlen()s saved. [ 27050] Yet more instances of gv_fetchpv... that should be GV_ADD rather than TRUE. Convert two gv_fetchpvn_flags to gv_fetchpvs. [ 27053] gv_fetchpvn_flags ranks highly in the profile, and the ::/' scanning loop is iterated over millions of times. Add a flag GV_NOTQUAL purely as an optimisation, when the caller is passing in a string that is known not to contain any package separators. Branch: maint-5.8/perl !> doio.c gv.c gv.h handy.h op.c perl.c pp_ctl.c pp_sort.c !> pp_sys.c toke.c util.c ____________________________________________________________________________ [ 29808] By: nicholas on 2007/01/14 13:22:20 Log: Integrate: [ 26946] SvREADONLY_off inside the magic routines is also dangerous on shared hash key scalars, so decompose them to normal scalars if they ever get here. [ 26954] Simplify nested '#ifdef' by replacing it with '#elif defined()' Branch: maint-5.8/perl !> mg.c ____________________________________________________________________________ [ 29807] By: nicholas on 2007/01/14 13:09:22 Log: Integrate: [ 26901] Change all NEWSV() to newSV() in the core and non-dual-lived modules. Keep NEWSV() itself for backwards-compatibility outside of the core, but don't advertise it any more. (cf. change #25101). [ 27455] Remove two NEWSV()s in the non-ithread dump code that got missed. Branch: maint-5.8/perl ! op.c perl.c pp_ctl.c util.c !> av.c doio.c dump.c ext/ByteLoader/bytecode.h !> ext/Opcode/Opcode.xs ext/POSIX/POSIX.xs gv.c handy.h hv.c !> mathoms.c os2/OS2/REXX/REXX.xs pad.c pod/perlapi.pod !> pod/perlembed.pod pp.c pp_hot.c pp_pack.c pp_sys.c regexec.c !> scope.c sv.c toke.c vms/ext/Stdio/Stdio.xs ____________________________________________________________________________ [ 29806] By: nicholas on 2007/01/14 12:49:30 Log: Integrate: [ 26765] Move initialization of old values prior to moreswitches() closer to their use and together with 'switches_done'. [ 26767] Missing an initialisation, as spotted by Merijn's HP compiler. [ 26786] Get rid of the following gcc format warnings by simplifying the getgroups implementation: mg.c: In function Perl_magic_get': mg.c:1008: warning: long unsigned int format, gid_t arg (arg 3) mg.c:1014: warning: long unsigned int format, gid_t arg (arg 3) mg.c:1025: warning: long unsigned int format, unsigned int arg (arg 3) Since we already cast the numeric Gid_t values to an IV it should not be too risky to also cast the Group_t values. Converting these values with Gid_t_f wasn't quite right anyway. [ 26787] Refactor S_vdie_common so that Perl_vwarn can use it too. [ 26791] Make $( and $) list the groups in the order they are returned from the OS. Linux seems to return the gids sorted and it seemed wrong for perl to reverse this order. [ 26893] Suppress "statement not reached" warning from the Sun C compiler. Branch: maint-5.8/perl ! mathoms.c proto.h !> doio.c embed.fnc embed.h mg.c toke.c util.c ____________________________________________________________________________ [ 29804] By: nicholas on 2007/01/14 12:22:41 Log: Integrate: [ 26643] Fix some compilation warnings [ 26674] Subject: [PATCH] performance tweaking op.c From: Andy Lester Date: Fri, 6 Jan 2006 01:44:48 -0600 Message-ID: <20060106074448.GB3401@petdance.com> [ 26708] Subject: [PATCH] blead: format warnings Date: Sat, 07 Jan 2006 21:42:08 +0200 From: Jarkko Hietaniemi Message-ID: <43C01990.6020207@gmail.com> [ 26764] Subject: [PATCH] It's the Barbie bus patch From: Andy Lester Date: Mon, 9 Jan 2006 23:42:43 -0600 Message-ID: <20060110054243.GA26165@petdance.com> Branch: maint-5.8/perl ! hv.c op.c regcomp.c !> av.c doio.c embed.fnc embed.h gv.c mg.c pad.c pp_ctl.c !> pp_hot.c pp_sys.c proto.h reentr.c reentr.h regexec.c sv.c !> toke.c uconfig.sh util.c ____________________________________________________________________________ [ 29802] By: nicholas on 2007/01/14 00:36:51 Log: Integrate: [ 23773] If we know the length of the string constant, save a strlen() [ 25172] Subject: [PATCH] The continuing plod through embed.fnc From: Andy Lester Date: Mon, 18 Jul 2005 10:37:38 -0500 Message-ID: <20050718153738.GB20193@petdance.com> [ 26635] Introduce STR_WITH_LEN macro in the form suggested by Chip. [ 26636] Fix typo in comment. [ 26640] Removed unused part of string passed to sv_catpvn(). [ 26641] Introduce the macros newSVpvs(str) and sv_catpvs(sv, str). Gets rid of many hardcoded string lengths. [ 26642] Get rid of a few more hardcoded string lengths. [ 26645] Get rid of a few more hardcoded string lengths. [ 26646] Introduce newSVpvs_share() macro. Gets rid of some hardcoded string lengths. [ 26647] Get rid of hardcoded string length. [ 26648] Add an extra NUL so that we can use sv_catpvs() for "our @F=split..." and get rid of 15 and the explaination for why it's there. [ 26649] Make the new STR_WITH_LEN() affected compile under -Dusethreads. Can't use STR_WITH_LEN() as argument to a macro :-( [ 26650] Move all the xxxpvs() macros to handy.h. This brings them closer to STR_WITH_LEN. Besides only half of them were sv-related after change 26649. [ 26671] Replacing broken call to savepvn() with savepvs() to get threaded Cygwin to compile again. [ 26676] Subject: [PATCH] hv_fetchs() support From: Andy Lester Date: Fri, 6 Jan 2006 09:57:19 -0600 Message-ID: <20060106155719.GB9035@petdance.com> [ 26795] A few more places that can use hv_fetchs(). Ref change 26676. Branch: maint-5.8/perl ! op.c !> doop.c dump.c ext/POSIX/POSIX.xs gv.c handy.h hv.c locale.c !> malloc.c mg.c perl.c perlio.c pp.c pp_ctl.c pp_pack.c pp_sys.c !> regcomp.c sv.c sv.h taint.c thread.h toke.c universal.c utf8.c !> util.c xsutils.c ____________________________________________________________________________ [ 29801] By: nicholas on 2007/01/13 23:40:49 Log: Integrate: [ 26598] silence some compiler warnings [ 26608] Fix release mode builds following change 26598 [ 26633] Silence compiler warning; ref change 26598 Branch: maint-5.8/perl ! cop.h !> op.c toke.c ____________________________________________________________________________ [ 29800] By: nicholas on 2007/01/13 23:25:42 Log: Integrate: [ 26562] Update copyrights for files modified in 2006 [ 26564] Subject: [patch] comment grammar and indentation, spurious tab From: Jim Cromie Date: Fri, 30 Dec 2005 22:45:55 -0700 Message-ID: <43B61B13.3050704@gmail.com> [ 26652] More copyright updates [ 26732] Update copyright years (including some years where we made changes but did not update) Branch: maint-5.8/perl ! sv.c !> cop.h deb.c doio.c dump.c hv.c locale.c mathoms.c mg.c mg.h !> perl.h perlio.c perlvars.h pp_ctl.c pp_pack.c pp_sort.c run.c !> scope.c sv.h taint.c thread.h ____________________________________________________________________________ [ 29799] By: nicholas on 2007/01/13 23:25:00 Log: Integrate: [ 26602] Bad symbols that are pretending to be dirhandles, should say they are dirhandles in their error messages. [ 26607] perldiag.pod additions to go with change #26602. [ 26617] Attemting to readdir() something that isn't a dirhandle should cause a warning. [ 26631] Add warnings for the various other *dir() functions when attempted on invalid dirhandles. [ 26638] Don't warn about invalid dirhandles in DirHandle::DESTROY(). Branch: maint-5.8/perl !> gv.c lib/DirHandle.pm pod/perldiag.pod pp_sys.c ____________________________________________________________________________ [ 29798] By: nicholas on 2007/01/13 22:58:25 Log: Integrate: [ 26494] Don't try to store PL_sv_undef in pads. (For now) [ 26518] A GVs stash can be NULL, so don't call macros that assume otherwise without checking. Branch: maint-5.8/perl !> op.c pp.c ____________________________________________________________________________ [ 29797] By: nicholas on 2007/01/13 22:58:03 Log: Integrate: [ 26538] In the shared string table, if we add new entries to the head of the linked list chain, then we don't need as much bookkeeping in the loop. Branch: maint-5.8/perl !> hv.c ____________________________________________________________________________ [ 29795] By: nicholas on 2007/01/13 22:37:35 Log: Integrate: [ 26524] Subject: [PATCH] Fixing the SVREFCNT macro From: Andy Lester Date: Wed, 28 Dec 2005 13:30:51 -0600 Message-ID: <20051228193051.GD23207@petdance.com> Branch: maint-5.8/perl !> sv.h ____________________________________________________________________________ [ 29794] By: nicholas on 2007/01/13 19:26:17 Log: Integrate: [ 26532] Subject: [PATCH] Perl_gv_fetchpv vs. gv_fetchpvn From: Tels Date: Thu, 29 Dec 2005 18:32:10 +0100 Message-Id: <200512291832.20651@bloodgate.com> [ 26533] Subject: [PATCH] perlio.c - gv_stashpvn() vs gv_stashpv() From: Tels Date: Thu, 29 Dec 2005 18:57:43 +0100 Message-Id: <200512291857.59458@bloodgate.com> [ 26556] Subject: [PATCH] Making 0 pointers to NULLs From: Andy Lester Date: Fri, 30 Dec 2005 23:05:58 -0600 Message-ID: <20051231050558.GA29093@petdance.com> [ 26558] Subject: [PATCH] Making 0 pointers to NULLs From: Andy Lester Date: Fri, 30 Dec 2005 23:05:58 -0600 Message-ID: <20051231050558.GA29093@petdance.com> Branch: maint-5.8/perl !> doio.c doop.c gv.c hv.c mg.c perlio.c pp.c pp_ctl.c pp_hot.c !> pp_sort.c regcomp.c regexec.c sv.c toke.c utf8.c util.c ____________________________________________________________________________ [ 29793] By: nicholas on 2007/01/13 18:59:04 Log: Integrate: [ 26527] Subject: [PATCH] Actual real consting of parms From: Andy Lester Date: Thu, 29 Dec 2005 01:15:09 -0600 Message-ID: <20051229071509.GA32653@petdance.com> Branch: maint-5.8/perl ! sv.c !> proto.h ____________________________________________________________________________ [ 29792] By: nicholas on 2007/01/13 18:44:35 Log: Integrate: [ 26526] Subject: [PATCH] av.c cleanup, plus! From: Andy Lester Date: Wed, 28 Dec 2005 15:44:14 -0600 Message-ID: <20051228214414.GD26033@petdance.com> Branch: maint-5.8/perl ! sv.c !> av.c dump.c gv.c perl.c toke.c ____________________________________________________________________________ [ 29791] By: nicholas on 2007/01/13 18:27:36 Log: Integrate: [ 26505] Subject: [PATCH] pre-likely cleanup From: Andy Lester Date: Tue, 27 Dec 2005 14:39:39 -0600 Message-ID: <20051227203939.GC1781@petdance.com> Includes a small fix to the changes in tryAMAGICbinW_var() in pp.h. Branch: maint-5.8/perl ! toke.c !> av.c doop.c hv.c mg.c pp.c pp.h sv.c util.c ____________________________________________________________________________ [ 29789] By: nicholas on 2007/01/13 18:10:34 Log: Integrate: [ 26461] Subject: [PATCH] Speed up Perl_sv_derived_from From: Andy Lester Message-ID: <20051222220044.GH4370@petdance.com> Date: Thu, 22 Dec 2005 16:00:44 -0600 [ 26477] For the rare case of EMFILE during require, save object code space by using newSVpvf (Would it be worth using it for all require failures?) [ 26500] Refactor the require failure message generation to use 1 less temporary SV, and to build as much text as possible with newSVpvf [ 26503] gv_fetch_flags in newATTRSUB can actually be const. Comment the logic behind the terms in the ternary that chooses the flags value. [ 26517] As the backreferences AV doesn't hold references on its contents, surely it should have AvREAL turned off? [ 26534] Reduce size of buffers for identifier names, as suggested in Subject: Stack usage (in gv_stashpvn and others) From: Tels Date: Thu, 29 Dec 2005 18:40:49 +0100 Message-Id: <200512291840.50765@bloodgate.com> [ 26541] Now that the backreference array has no NULL entries, and is correctly marked as AvREAL_off(), there's no reason not to use sv_dup to clone it. [ 26544] Tweak S_init_main_stash so as allocate PL_curstname as a shared string scalar, and hence avoid thrashing the shared string table for "main". [ 26550] Remove unneeded test in Perl_hv_clear_placeholders. Rejig S_hfreeentries to a double loop, which is clearer and smaller. [ 26552] Turn a for loop that's almost a while into an honest-to-goodness while. Branch: maint-5.8/perl ! gv.c hv.c !> op.c perl.c pp_ctl.c sv.c universal.c ____________________________________________________________________________ [ 29788] By: nicholas on 2007/01/13 17:39:10 Log: Integrate: [ 26456] Subject: [PATCH] Speedup (rn|n)?instr From: Andy Lester Message-ID: <20051222223839.GI4370@petdance.com> Date: Thu, 22 Dec 2005 16:38:39 -0600 [ 26509] Fix 2 off-by-one errors in the call to ninstr(). This code managed to still work because of a bug in how ninstr() treats empty search strings. [ 26510] Remove test for NUL ending in r?instr(). This test is actually harmful because we will search for any garbage found past the empty search string. Ref change 26509. [ 26511] Leaner ninstr(). The compiled function ended up 37% smaller on Linux/gcc. I ought to be faster as well, but I did not try to measure that. [ 26512] Use memchr() instead of ninstr() to locate the end-of-line. The call to memchr() will faster. Branch: maint-5.8/perl !> ext/Filter/Util/Call/Call.xs toke.c util.c ____________________________________________________________________________ [ 29787] By: nicholas on 2007/01/13 17:24:48 Log: Integrate: [ 26450] Subject: [PATCH] Ho! Ho! Ho! Santa brings consting! From: Andy Lester Date: Thu, 22 Dec 2005 10:57:17 -0600 Message-ID: <20051222165717.GA2874@petdance.com> Branch: maint-5.8/perl ! global.sym proto.h !> dump.c embed.fnc embed.h gv.c hv.c op.c pp.c pp_ctl.c pp_hot.c !> pp_sys.c regcomp.c sv.c toke.c universal.c ____________________________________________________________________________ [ 29785] By: nicholas on 2007/01/13 16:56:44 Log: Integrate: [ 26446] Add an optimisation to allow proxy constant subroutines to be copied as proxy constant subroutines in a new symbol table where possible. (Rather than converting them to full blown constant subroutines and instantiating 2 typeglobs) [ 26448] Regression tests for proxy subroutine glob assignment. Fix a bug (it turns out that a typeglob isn't SvOK()) Remove stray debugging code. [ 26470] Now that proxy subroutines can be unproxied in 2 places without becoming the same newCONSTSUB, need an explicit check to avoid a warning about subroutines being redefined, as there has never been a warning when you assing the same subroutine to a glob's GvCV() [ 26482] Unless the peephole optimiser already knows that we're in void context, avoid even attempting the whole proxy constant sub copying optimisation. Branch: maint-5.8/perl ! op.c !> ext/B/t/concise-xs.t gv.c gv.h op.h pp.c pp_hot.c sv.c !> t/op/gv.t ____________________________________________________________________________ [ 29783] By: nicholas on 2007/01/13 16:30:48 Log: Integrate: [ 26409] Make gv_init recognise a reference-to-something in a symbol table as meaning a constant subroutine with that thing as it value [ 26417] Verify that the code for initialising typeglobs from other types works. [ 26422] Croak if gv_init doesn't know how to create a typeglob from that type of referant. Test that ARRAY, HASH, PVIO, CODE and FORMAT croak. Globs are actually first class assignable objects, so test that you can create a constant subroutine that returns one. [ 26425] Add a GV_NOADD_NOINIT flag to gv_fetch{pv,pvn,sv} that disables addition of new typeglobs, and also disables initialisation of any typeglob placeholders. Needed to make the new constant subroutine proxy references work efficiently. [ 26427] Take care in toke.c not to convert constant subroutine reference proxies into full blown PVGVs with PVCVs, and recognise them and inline their values. Adds a new function gv_const_sv(gv) to return the value of the constant subroutine from a GV, and adds a cv parameter to S_intuit_method. Branch: maint-5.8/perl ! toke.c !> embed.fnc embed.h gv.c gv.h pod/perldiag.pod proto.h t/op/gv.t ____________________________________________________________________________ [ 29782] By: nicholas on 2007/01/13 13:47:04 Log: Integrate: [ 26411] Assert that orig_keyword, gv and gvp aren't used before the main keyword parser of Perl_yylex. [ 26412] In Perl_yylex, make tmp a temporary variable local to the blocks in which it is used. [ 26414] In Perl_yylex, move the declaration of orig_keyword, gv and gvp down to the block labelled keylookup. [ 26423] Get the "cv" from the "gv" once, and be more careful so that we can cope if either aren't their regular types. cv_const_sv should verify the type of SV passed, instead of assuming that it's a viable CV. Branch: maint-5.8/perl ! toke.c !> op.c ____________________________________________________________________________ [ 29776] By: nicholas on 2007/01/12 14:24:33 Log: Integrate: [ 26404] Make sv_dump (and therefore Devel::Peek) report the value of the constant in inlineable constant subroutines. Branch: maint-5.8/perl !> dump.c ext/Devel/Peek/t/Peek.t ____________________________________________________________________________ [ 29775] By: nicholas on 2007/01/12 14:07:47 Log: Integrate: [ 26334] A suggestion by Yamashina Hio to speed up substitutions with right-hand side expressions by freeing temporaries. See : Subject: s///ge; consumes PL_tmps_stack in its loop From: YAMASHINA Hio Date: Tue, 30 Aug 2005 17:17:23 +0900 Message-Id: <20050830160113.9716.HIO@ymir.co.jp> [ 26369] In Perl_apply, the name of the op can be found from PL_op_name, instead of using local string constants. [ 26372] Use PL_op_desc rather than literal strings in pp_ioctl [ 26392] Move and const some definitions of variables used to save errno. [ 26437] Localise the stash variable in pp_entersub(). [ 26438] For consistency sv_2cv should not leave the stash pointer uninitialised for some cases. Although it seems that only pp_sort pays any attention to it, of all the code on CPAN. [ 26444] Remove unused variable 'stash' since change 26437. Branch: maint-5.8/perl !> doio.c perlio.c pp_ctl.c pp_hot.c pp_sys.c sv.c ____________________________________________________________________________ [ 29773] By: nicholas on 2007/01/12 13:29:41 Log: Integrate: [ 26301] Merge common code from ptr_table_fetch and ptr_table_store into S_ptr_table_find. [ 26310] Fixes compile errors introduce with change #26301 when compiling with ithreads and DEBUGGING. [ 26312] A more efficient way to loop in ptr_table_clear [ 26313] An alternative way of structuring ptr_table_clear so that the variables can be assigned to at declaration time. [ 26317] Subject: [PATCH] Code tweaks in sv.c From: Andy Lester Date: Fri, 9 Dec 2005 11:09:21 -0600 Message-ID: <20051209170921.GC16943@petdance.com> Includes a minor change the patch to account for code that moved elsewhere. Branch: maint-5.8/perl !> sv.c ____________________________________________________________________________ [ 29772] By: nicholas on 2007/01/12 13:28:49 Log: Integrate: [ 26360] Add a missing FREETMPS to match up with a SAVETMPS. This change fixes problems with nested sorts. (RTs #7063 and #36430). [ 26361] Again, FREETMPS not SAVETMPS! [ 26362] Reverting change #26360 and #26361 since it appears that the bugs I thought I fixed had been previously fixed with change #25953. Branch: maint-5.8/perl !> pp_sort.c ____________________________________________________________________________ [ 29771] By: nicholas on 2007/01/12 13:01:39 Log: Integrate: [ 26281] Subject: [PATCH] More consting, and putting stuff in embed.fnc From: Andy Lester Date: Mon, 5 Dec 2005 13:46:13 -0600 Message-ID: <20051205194613.GB7791@petdance.com> [ 26287] Add a ; that 26281 was lacking. (Dear Mark, please can I have back one of the two that I sent you after change 23614) Branch: maint-5.8/perl !> deb.c doio.c dump.c embed.fnc embed.h hv.c hv.h locale.c mg.c !> op.c op.h pp.c pp_ctl.c pp_hot.c pp_pack.c pp_sort.c pp_sys.c !> proto.h regcomp.c sv.c taint.c toke.c utf8.c util.c ____________________________________________________________________________ [ 29770] By: nicholas on 2007/01/12 12:22:53 Log: Integrate: [ 26199] Subject: XS-assisted SWASHGET (esp. for t/uni/class.t speedup) From: SADAHIRO Tomoyuki Date: Wed, 23 Nov 2005 17:57:34 +0900 Message-Id: <20051123175603.FFD5.BQW10602@nifty.com> And : Message-Id: <20051123202935.4D9D.BQW10602@nifty.com> with some nits to use U8 instead of char more consistently [ 26203] Subject: Re: XS-assisted SWASHGET (esp. for t/uni/class.t speedup) From: SADAHIRO Tomoyuki Date: Fri, 25 Nov 2005 00:19:35 +0900 Message-Id: <20051125001031.24E3.BQW10602@nifty.com> Date: Fri, 25 Nov 2005 01:18:17 +0900 Message-Id: <20051125011410.24E9.BQW10602@nifty.com> [ 26229] Subject: Re: XS-assisted SWASHGET (esp. for t/uni/class.t speedup) From: SADAHIRO Tomoyuki Date: Sun, 27 Nov 2005 17:02:02 +0900 Message-Id: <20051127170016.A786.BQW10602@nifty.com> [ 26255] Clarification and cleanup of the XS SWASHGET code Subject: Re: XS-assisted SWASHGET (esp. for t/uni/class.t speedup) From: SADAHIRO Tomoyuki Date: Sun, 04 Dec 2005 16:28:35 +0900 Message-Id: <20051204162508.D726.BQW10602@nifty.com> [ 26256] Rework the error messages from the swashget code. Subject: Re: XS-assisted SWASHGET (esp. for t/uni/class.t speedup) From: SADAHIRO Tomoyuki Date: Sun, 04 Dec 2005 16:24:59 +0900 Message-Id: <20051204162431.D723.BQW10602@nifty.com> [ 26257] Better variable names make code clearer Subject: Re: XS-assisted SWASHGET (esp. for t/uni/class.t speedup) From: SADAHIRO Tomoyuki Date: Sun, 04 Dec 2005 16:24:29 +0900 Message-Id: <20051204162412.D720.BQW10602@nifty.com> Branch: maint-5.8/perl !> embed.fnc embed.h lib/utf8_heavy.pl pod/perlapi.pod !> pod/perldiag.pod pod/perlintern.pod proto.h regexec.c !> t/op/pat.t universal.c utf8.c utf8.h utfebcdic.h ____________________________________________________________________________ [ 29769] By: nicholas on 2007/01/12 11:55:51 Log: Integrate: [ 26090] const'ing Netdb_host_t broke HP-UX and created warnings on gcc for a double const. Netdb_host_t is a macro that for a const pointer. [ 26147] Subject: [PATCH] Sort subs now in embed.fnc From: Andy Lester Date: Thu, 17 Nov 2005 07:53:38 -0600 Message-ID: <20051117135338.GA31715@petdance.com> [ 26175] Cleaning up some warnings generated by "gcc -W" [ 26179] Declarations must come first, or some compilers aren't happy. (spotted by Jarkko) Branch: maint-5.8/perl !> deb.c embed.fnc embed.h ext/B/B.xs ext/Data/Dumper/Dumper.xs !> ext/PerlIO/encoding/encoding.xs mg.c op.c pp_sort.c pp_sys.c !> proto.h regcomp.c scope.c ____________________________________________________________________________ [ 29763] By: nicholas on 2007/01/12 01:57:13 Log: Integrate: [ 25956] Adjust test count Branch: maint-5.8/perl !> ext/B/t/concise-xs.t ____________________________________________________________________________ [ 29760] By: nicholas on 2007/01/11 17:10:41 Log: Integrate: [ 26082] Subject: [PATCH] Const & local: Special Victims Unit From: Andy Lester Date: Thu, 10 Nov 2005 22:47:00 -0600 Message-ID: <20051111044700.GA24161@petdance.com> Branch: maint-5.8/perl !> cop.h doio.c doop.c pp_hot.c pp_pack.c pp_sort.c pp_sys.c sv.c !> util.c ____________________________________________________________________________ [ 29759] By: nicholas on 2007/01/11 16:45:05 Log: Integrate: [ 25953] Subject: [PATCH] sort/multicall patch From: Robin Houston Date: Sat, 29 Oct 2005 21:33:07 +0100 Message-ID: <20051029203307.GA8869@rpc142.cs.man.ac.uk> [ 25954] Regen headers [ 25957] Remove last sort test, that was failing with and without threads. Silence some warnings. [ 25979] Fix test failures introduced by the change of flags on op_sort [ 25992] Subject: Re: [PATCH] sort/multicall patch From: Robin Houston Date: Fri, 4 Nov 2005 15:20:29 +0000 Message-ID: <20051104152029.GA17169@rpc142.cs.man.ac.uk> [ 26054] Subject: [PATCH] Re: [perl #32383] DProf breaks List::Util::shuffle From: Robin Houston Date: Tue, 8 Nov 2005 19:02:34 +0000 Message-ID: <20051108190234.GA25953@rpc142.cs.man.ac.uk> [ 27486] Subject: [PATCH] Automatically SPAGAIN in the poor-man's-multicall API From: Robin Houston Date: Mon, 13 Mar 2006 12:48:21 +0000 Message-ID: <20060313124821.GA11203@rpc142.cs.man.ac.uk> [ 27487] Forgot the ending \. Why did it still compile ? Branch: maint-5.8/perl ! cop.h embed.fnc pp_sort.c proto.h utf8.c !> AUTHORS embedvar.h ext/B/t/f_sort.t ext/List/Util/multicall.h !> global.sym makedef.pl op.c opcode.h opcode.pl perlapi.h !> pod/perlapi.pod pod/perlcall.pod pod/perldiag.pod pp_ctl.c !> pp_hot.c sv.c t/op/sort.t ____________________________________________________________________________ [ 29755] By: nicholas on 2007/01/11 14:29:40 Log: Integrate: [ 26028] Subject: [PATCH] Consting and localizing: Part LXVIII From: Andy Lester Date: Fri, 4 Nov 2005 15:12:56 -0600 Message-ID: <20051104211256.GA12651@petdance.com> Branch: maint-5.8/perl !> deb.c embed.fnc embed.h locale.c pp_sys.c scope.c universal.c !> utf8.c util.c ____________________________________________________________________________ [ 29754] By: nicholas on 2007/01/11 13:30:43 Log: Integrate: [ 25941] Subject: futimes [PATCH] From: Gisle Aas Date: 31 Oct 2005 13:53:53 -0800 Message-ID: [ 25951] Subject: Re: futimes [PATCH] From: Gisle Aas Date: 01 Nov 2005 00:02:37 -0800 Message-ID: Branch: maint-5.8/perl !> doio.c handy.h pod/perlfunc.pod t/io/fs.t ____________________________________________________________________________ [ 29752] By: nicholas on 2007/01/11 12:41:42 Log: Integrate: [ 25915] Subject: [PATCH] Etta James presents: More consting From: Andy Lester Date: Sun, 30 Oct 2005 23:44:13 -0600 Message-ID: <20051031054413.GA10767@petdance.com> Branch: maint-5.8/perl ! universal.c !> cop.h doop.c hv.c numeric.c op.c perl.c pp_hot.c pp_sys.c sv.c !> toke.c util.c ____________________________________________________________________________ [ 29746] By: nicholas on 2007/01/10 15:52:19 Log: Integrate: [ 29729] socketpair.t was written in 2001. At that time, Test::More got confused if you forked under it. Test::More 0.47 added support to detect forking which resolves this. Hence there is now no need to fork() in a BEGIN block, which causes issues on Win32 due to how pseudofork works. Branch: maint-5.8/perl !> ext/Socket/t/socketpair.t ____________________________________________________________________________ [ 29737] By: nicholas on 2007/01/09 21:23:37 Log: Conditionally compile out all op.c manipulation code for OP_THREADSV unless we are building for 5005threads, as nothing else can generate this op. (Arguably pre 5.8.0 time we could have changed the headers to not even have this op usually, but it's a tad late now and my time machine is still on pre-order). Branch: maint-5.8/perl ! op.c ____________________________________________________________________________ [ 29735] By: nicholas on 2007/01/09 12:48:39 Log: We can get in the messy situation of the COP that PL_curcop pointed to getting freed, and as part of the same free overloading decides to look for DESTROY, which needs to *create* a GV, which in turn was expecting that PL_curcop pointed to something valid. So set PL_curcop to NULL if we're freeing the COP that it points to, and make Perl_gv_init() cope with a NULL PL_curcop. Branch: maint-5.8/perl ! gv.c op.c ____________________________________________________________________________ [ 29733] By: nicholas on 2007/01/09 11:12:58 Log: Integrate: [ 25973] More core test cleanups. Mainly fixes to remove code that caused warnings, but there were a few bug fixes that were hidden by no using warnings. [ 26066] I/O is UTF-8, not "UNICODE" [ 26070] Add tests for two untested syntax error messages about \N [ 26178] Subject: [PATCH] Pulling ancient RCS comments From: Andy Lester Date: Sat, 19 Nov 2005 00:16:39 -0600 Message-ID: <20051119061639.GA25086@petdance.com> [ 26347] Convert range.t to test.pl Branch: maint-5.8/perl !> (integrate 36 files) ____________________________________________________________________________ [ 29726] By: nicholas on 2007/01/08 23:25:40 Log: Unbreak 5005 threads - pp_threadsv can't be aliased to unimplemented_op for the 5005 threads case. Branch: maint-5.8/perl ! opcode.h opcode.pl ____________________________________________________________________________ [ 29725] By: nicholas on 2007/01/08 22:09:41 Log: Integrate: [ 29695] Update copyright years in .h files. Also, in .pl files that generate .h files, so they'll be ready next time. [just the .pl files for now] Branch: maint-5.8/perl !> keywords.pl opcode.pl ____________________________________________________________________________ [ 29722] By: nicholas on 2007/01/08 18:06:13 Log: Integrate: [ 26110] Integrate: [ 26108] BEGIN blocks should start a new stack, as they can be called as a side effect of "regular" Perl API calls within subroutines that have already cached the current stack's address. If they don't, any stack extension during the call may move the stack, rendering that cached value invalid without the cachee realising. (For example, PP code calling gv_fetchpv() which triggers a load of Errno.pm) [ 26234] Revert change #26110 (already reverted in maintperl by change 26182) [ 27207] Subject: Re: Perl_sighandler voodoo (was Re: Smoke [5.8.7] 26127 FAIL(XF) MSWin32 WinXP/.Net SP2 (x86/2 cpu)) From: Robin Houston Date: Sun, 27 Nov 2005 00:43:13 +0000 Message-ID: <20051127004313.GA7007@rpc142.cs.man.ac.uk> Simplification of the code in my_exit_jump() that unwinds context stacks Branch: maint-5.8/perl ! t/io/open.t !> op.c perl.c ____________________________________________________________________________ [ 29721] By: nicholas on 2007/01/08 15:51:22 Log: Integrate: [ 26126] Don't warn about lack of docs for functions that appear twice in embed.fnc. [ 29138] Modify autodoc.pl to add read-only editor boilerplate for perlapi.pod and perlguts.pod [ 29146] In vim, modelines are not taken into account if they begin on the first character. Also, regenerate perlintern.pod. Branch: maint-5.8/perl !> autodoc.pl pod/perlapi.pod pod/perlguts.pod pod/perlintern.pod ____________________________________________________________________________ [ 29720] By: nicholas on 2007/01/08 15:43:25 Log: Integrate: [ 25703] Add -C to allowed $PERL5OPT options Add -C to the $PERL5OPT documentation (and added the missing -A too) t/run/runenv.t Could use some more tests about how supported and unsupported options in $PERL5OPT are handled Branch: maint-5.8/perl ! pod/perldiag.pod pod/perlrun.pod !> perl.c ____________________________________________________________________________ [ 29163] By: nicholas on 2006/10/30 18:26:48 Log: Integrate: [ 26177] Subject: Reworked PERL_TRACK_MEMPOOL patch From: "Jan Dubois" Date: Fri, 18 Nov 2005 11:38:24 -0800 Message-ID: <003601c5ec77$a45eb260$2217a8c0@candy> [tweaked so as to remove the automatic define of -DPERL_TRACK_MEMPOOL on -DDEBUGGING] Branch: maint-5.8/perl !> perl.h util.c ____________________________________________________________________________ [ 29141] By: nicholas on 2006/10/29 22:02:32 Log: Integrate: [The Perl_magic_killbackrefs change from:] [ 24966] Remove the reference loop between symbol tables and typeglobs. Typeglobs now have a weak reference onto their symbol table. [ 26132] Mark all places where perl needs to look at a possibly-freed scalar with a macro SvIS_FREED(sv) Branch: maint-5.8/perl !> av.c mg.c op.c pad.c pp_hot.c sv.h ____________________________________________________________________________ [ 29137] By: nicholas on 2006/10/29 20:23:57 Log: Integrate: [ 26060] Add a comment about possible future refactoring of the pad code. [ 26123] Suppress gcc warnings when ignoring the return value of io_close(), as suggested by Gisle Aas. [ 26124] Fix indentation of apidoc for sv_2pvutf8 (so perlapi.pod is correctly generated) [ 26176] Down with "warning: code has no effect" Branch: maint-5.8/perl !> XSUB.h embed.fnc pad.c pod/perlapi.pod proto.h sv.c ____________________________________________________________________________ [ 29136] By: nicholas on 2006/10/29 20:01:41 Log: Integrate: [ 26112] Fix breakages that prevended -DPERL_POISON from compiling. Branch: maint-5.8/perl !> handy.h sv.h ____________________________________________________________________________ [ 29135] By: nicholas on 2006/10/29 19:31:39 Log: Integrate: [ 25917] POD nit spotted by Jarkko [ 25938] Further pod nit spotted by Jarkko [ 25944] Fix nit in pod nit fix Branch: maint-5.8/perl !> pod/perlhack.pod ____________________________________________________________________________ [ 29134] By: nicholas on 2006/10/29 19:28:47 Log: Integrate: [ 25914] Subject: [PATCH] S_tokereport's unused parm From: Andy Lester Date: Mon, 31 Oct 2005 00:04:59 -0600 Message-ID: <20051031060459.GC10767@petdance.com> Branch: maint-5.8/perl !> embed.fnc embed.h proto.h toke.c ____________________________________________________________________________ [ 29132] By: nicholas on 2006/10/29 19:16:37 Log: Integrate: [ 25808] Subject: Re: [PATCH] Re: [perl #37350] $#{@$aref} in debugger gives: Bizarre copy of ARRAY in leave From: Robin Houston Date: Oct 14, 2005 1:54 AM Message-ID: <20051013235457.GA23386@rpc142.cs.man.ac.uk> Branch: maint-5.8/perl ! mathoms.c op.c t/op/array.t !> embed.fnc embed.h global.sym op.h proto.h ____________________________________________________________________________ [ 29129] By: nicholas on 2006/10/29 18:29:46 Log: Integrate: [ 25697] Subject: proposal [perl #34301]: IO::Socket calls getpeername far too often From: Peter.Dintelmann@Dresdner-Bank.com (Peter Dintelmann) Date: Thu, 2 Jun 2005 13:34:49 +0200 Message-ID: [ 25861] Subject: Re: IO::File performace issue From: gisle@ActiveState.com (Gisle Aas) Date: 27 Oct 2005 01:00:39 -0700 Message-ID: Also includes a version bump for IO::File. [ 26300] Fixup the setsockopt usage message; missing OPTVAL. [ 26400] Subject: [PATCH] latest switch/say/~~ From: Robin Houston Date: Sat, 17 Dec 2005 20:44:31 +0000 Message-Id: <20051217204431.GB28940@rpc142.cs.man.ac.uk> [ 26405] Attempt at making IO::Handle backwards compatible again [ 26481] Given that IO::Socket is documented as exporting all of Socket's exports, there's no need for IO::Socket::INET or IO::Socket::UNIX to use both. (And re-exporting removes some of the efficiency of the new constants implementation) [ 26868] Don't warn about a directory being closed in the DESTROY for IO::Dir. [ 26921] Subject: Avoid most getprotobyname/number calls in IO::Socket::INET From: Gisle Aas Date: 18 Jan 2006 09:28:24 -0800 Message-ID: [ 26941] Subject: Simplify &IO::Handle::say From: Gisle Aas Date: 24 Jan 2006 05:01:28 -0800 Message-ID: [ 27609] Update to IO-1.23 Branch: maint-5.8/perl !> ext/IO/ChangeLog ext/IO/IO.pm ext/IO/lib/IO/Dir.pm !> ext/IO/lib/IO/File.pm ext/IO/lib/IO/Handle.pm !> ext/IO/lib/IO/Socket.pm ext/IO/lib/IO/Socket/INET.pm !> ext/IO/lib/IO/Socket/UNIX.pm ext/IO/t/io_multihomed.t !> ext/IO/t/io_pipe.t ext/IO/t/io_sock.t ext/IO/t/io_taint.t !> ext/IO/t/io_tell.t ext/IO/t/io_udp.t ext/IO/t/io_unix.t ____________________________________________________________________________ [ 29087] By: nicholas on 2006/10/22 23:09:33 Log: Integrate: [ 27737] Give more information in the panic in ss_dup. (Although I'm not sure if croak() works at this point, given PerlIO's dependency on a functional interpreter.) [ 29084] Fix a bug introduced by change 29079 [ 29086] Fix the other bug introduced by change 29079 Branch: maint-5.8/perl !> sv.c ____________________________________________________________________________ [ 29081] By: nicholas on 2006/10/22 21:00:37 Log: No need to save the PVX to the save stack when saving a GV, as the string buffer is only used as part of get/set magic when reading/ writing the stringified form. Keep the old save type in case any code out there is still generating the old format on the save stack. Branch: maint-5.8/perl ! scope.c scope.h sv.c ____________________________________________________________________________ [ 29080] By: nicholas on 2006/10/22 20:12:38 Log: Integrate: [ 29079] const the variable used for the save stack type in Perl_ss_dup(). Branch: maint-5.8/perl !> sv.c ____________________________________________________________________________ [ 29022] By: nicholas on 2006/10/15 09:41:52 Log: Integrate: [ 25254] Another typo fix Branch: maint-5.8/perl !> XSUB.h pod/perlapi.pod ____________________________________________________________________________ [ 29021] By: nicholas on 2006/10/15 09:22:52 Log: Integrate: [ 27987] Subject: Re: [PATCH] use snprintf/strlcpy/strlcat when useful From: Jarkko Hietaniemi Date: Tue, 25 Apr 2006 18:23:39 +0300 Message-ID: <444E3EFB.8020503@gmail.com> [ 27988] Fix typo in function call Branch: maint-5.8/perl !> op.c perl.h perlio.c pp_ctl.c regcomp.c sv.c toke.c !> universal.c util.c ____________________________________________________________________________ [ 29014] By: nicholas on 2006/10/14 16:42:04 Log: Integrate: [ 25105] Message-ID: <42D0F25E.3040801@gmail.com> Date: Sun, 10 Jul 2005 13:03:10 +0300 From: Jarkko Hietaniemi Subject: [PATCH] yet another way of debugging memory allocations adds PERL_MEM_LOG and PERL_MEM_LOG_STDERR options [ 25109] From: Jarkko Hietaniemi Date: Sun, 10 Jul 2005 21:50:27 +0300 Subject: [PATCH] yet another way of debugging memory allocations Message-ID: <42D16DF3.4040806@gmail.com> tweak PERL_MEM_LOG [ 25115] Subject: [patch: handy.c] update Newx API pod to mention PERL_MEM_LOG build opt From: Jim Cromie Date: Mon, 11 Jul 2005 19:18:02 -0600 Message-ID: <42D31A4A.1020305@divsol.com> (with some minor tweaks) [ 25116] Change New*() to Newx*() in various comments and documentation [ 25128] Subject: Make PERL_MEM_LOG more portable. From: Jarkko Hietaniemi Date: Mon, 11 Jul 2005 15:29:46 +0300 Message-Id: <42D2663A.4050204@gmail.com> [ 27204] Subject: Re: how to build with -DPERL_MEM_LOG ? From: Andy Dougherty Date: Fri, 3 Feb 2006 11:55:58 -0500 (EST) Message-ID: Branch: maint-5.8/perl !> handy.h pod/perlapi.pod pod/perlhack.pod util.c x2p/a2p.h ____________________________________________________________________________ [ 29013] By: nicholas on 2006/10/14 15:13:34 Log: Integrate: [ 25104] silence some compiler warnings Branch: maint-5.8/perl !> embed.fnc gv.h proto.h ____________________________________________________________________________ [ 28854] By: nicholas on 2006/09/16 17:19:07 Log: Integrate: [ 24803] hash key code cleanup Branch: maint-5.8/perl !> embed.fnc hv.c hv.h proto.h ____________________________________________________________________________ [ 28852] By: nicholas on 2006/09/16 16:20:26 Log: Integrate: [ 24352] Skip PL_sig_sv when PERL_IMPLICIT_CONTEXT is defined (rather than when PERL_IMPLICIT_SYS is defined) Branch: maint-5.8/perl !> makedef.pl ____________________________________________________________________________ [ 28775] By: nicholas on 2006/09/02 16:01:36 Log: Integrate: [ 28771] Change the generation of {} and [] from 3 ops to 1, and avoid 1 mortal on the tempstack, by augmenting pp_anonlist and pp_anonhash to accept OPf_SPECIAL to mean "return a reference to the aggregate" on the stack rather than the aggregate itself. Branch: maint-5.8/perl !> ext/B/B/Deparse.pm ext/B/t/concise-xs.t ext/B/t/f_map.t !> ext/B/t/f_sort.t ext/Devel/Peek/t/Peek.t op.c op.h pp.c ____________________________________________________________________________ [ 28443] By: nicholas on 2006/06/27 22:39:26 Log: Integrate: [ 28189] A Configure probe for C99 variadic macros, based on code from Jarkko. [ 28263] Oops. Fix bad #! lines when not using userelocatableinc [ 28283] Re-generation after backports of #28188 Only ask about MAD for 5.9.0 and greater. Otherwise silently #28189 A Configure probe for C99 variadic macros, based on code from #28202 Take advantage of the bug that binexp ignores the Configure #28263 Oops. Fix bad #! lines when not using userelocatableinc One small cleanup fix added. Some small re-ordering. Some whitespace. [ 28284] If I needed this now, I'll probably need it next time too :) [ 28286] HAS_HAS WAS NOW HAS Branch: maint-5.8/perl !> Configure Cross/config.sh-arm-linux NetWare/config.wc !> Porting/Glossary Porting/config.sh config_h.SH configure.com !> epoc/config.sh handy.h plan9/config_sh.sample uconfig.sh !> win32/config.bc win32/config.gc win32/config.vc !> win32/config.vc64 ____________________________________________________________________________ [ 28442] By: nicholas on 2006/06/27 22:20:20 Log: Integrate: [ 28024] OK, Win32 must no have snprintf()/vsnprintf [ 28025] I was mistaken. Borland and gcc do have snprintf() and vsnprintf() on Win32. [ 28026] Subject: [PATCH] fix 27987 for MSVC (was Re: Smoke [5.9.4] 28020 FAIL(m) MSWin32 Win2000 SP4 (x86/1 cpu)) From: Abe Timmerman Date: Sun, 30 Apr 2006 15:15:24 +0200 Message-Id: <200604301515.24580.abe@ztreet.demon.nl> [ 28051] Subject: RE: [PATCH] fix 27987 for MSVC (was Re: Smoke [5.9.4] 28020 FAIL(m) MSWin32 Win2000 SP4 (x86/1 cpu)) From: "Jan Dubois" Date: Mon, 1 May 2006 18:55:11 -0700 Message-ID: <08ff01c66d8b$7282e320$2217a8c0@candy> Subject: Re: [PATCH] fix 27987 for MSVC (was Re: Smoke [5.9.4] 28020 FAIL(m) From: "H.Merijn Brand" Date: Tue, 2 May 2006 09:23:51 +0200 Message-ID: <20060502092351.7bae0896@pc09> [ 28052] Ahem. And here's the rest of #28051... [ 28053] Sigh. Third time lucky. Branch: maint-5.8/perl !> win32/config.vc win32/config.vc64 win32/config_H.bc !> win32/config_H.gc win32/config_H.vc win32/config_H.vc64 !> win32/win32.h ____________________________________________________________________________ [ 28441] By: nicholas on 2006/06/27 22:15:42 Log: Integrate: [ 27863] Support for -DDEBUGGING and its alias -DEBUGGING Branch: maint-5.8/perl !> Configure INSTALL config_h.SH ____________________________________________________________________________ [ 28440] By: nicholas on 2006/06/27 22:00:05 Log: Integrate: [ 27686] First stab at a more reliable Glossary. This now is (again) mostly generated. (Well, It's a repair to the existing version "based on" a generated version, which is now well sorted, but still incomplete for less obvious reasons). Branch: maint-5.8/perl !> Porting/Glossary ____________________________________________________________________________ [ 28439] By: nicholas on 2006/06/27 21:54:10 Log: Integrate: [ 27370] Subject: [PATCH] Support for __builtin_expect and __builtin_choose_expr From: Andy Lester Date: Wed, 28 Dec 2005 15:34:08 -0600 Message-ID: <20051228213408.GB26033@petdance.com> [ 27372] Make Win32 configuration changes for HAS_BUILTIN_CHOOSE_EXPR and HAS_BUILTIN_EXPECT. [ 27373] Add d_builtin_choose_expr and d_builtin_expect to configure.com as 'undef'. [ 27412] Complete change #27407/27410 on Win32 and tidy up change #27372 [ 27473] Subject: [PATCH] Reapply change 24432 From: "Jan Dubois" Date: Fri, 10 Mar 2006 18:21:39 -0800 Message-ID: <01b001c644b2$879a1f50$6062a8c0@candy> Branch: maint-5.8/perl !> Configure Porting/Glossary config_h.SH configure.com perl.h !> win32/config.bc win32/config.gc win32/config.vc !> win32/config.vc64 win32/config_H.bc win32/config_H.gc !> win32/config_H.vc win32/config_H.vc64 ____________________________________________________________________________ [ 28438] By: nicholas on 2006/06/27 21:34:26 Log: Integrate: [ 27363] Subject: [perl #38385] _h2ph_pre.ph / $Config{cppsymbols} omits gcc-3.4+ cpp "predefined macros" Date: Tue, 31 Jan 2006 09:50:32 -0800 From: Jason Vas Dias (via RT) [ 27367] #27363 broke HP-UX (and others). if [[ ... ]] syntax not supported Branch: maint-5.8/perl !> Configure utils/h2ph.PL ____________________________________________________________________________ [ 28437] By: nicholas on 2006/06/27 21:21:12 Log: Integrate: [ 27189] Subject: Patch for perl to compile/work on DragonFlyBSD From: Robert Sebastian Gerus Date: Tue, 14 Feb 2006 17:27:52 +0100 Message-ID: <4a1e16420602140827w17fd6595w@mail.gmail.com> [ 27190] Added hints/dragonfly.sh for #27189 [ 27263] Additional hints needed to build threaded Perls on NetBSD. Branch: maint-5.8/perl +> hints/dragonfly.sh !> Configure MANIFEST Makefile.SH hints/netbsd.sh ____________________________________________________________________________ [ 28436] By: nicholas on 2006/06/27 20:49:08 Log: Integrate: [ 26606] Add probe for snprintf () and vsnprintf () Subject: [PATCH] snprintf() and vsnprintf() (was Re: [PATCH] Configure: stdio char (un)signedness) From: Steve Peters Date: Mon, 2 Jan 2006 09:37:27 -0600 Message-ID: <20060102153727.GA14247@mccoy.peters.homeunix.org> [ 26630] Add snprintf() and vsnprintf() to the various Win32 configuration files. [ 26632] Added snprintf() and vsnprintf() to configure.com as 'undef'. [ 26655] Reorder changes made in change 26630 (Keep the order in win32/config.* and win32/config_H.* the same as in Configure and config_H.SH for easier diffing) [ 26668] Test if the probed availability of v?snprintf () is actually usable. Thanks to Russ and Jarkko. [ 26703] Make v?snprintf available on VMS v7.3-2 and later and clean up some other version-dependency checks. Branch: maint-5.8/perl !> Configure Porting/Glossary config_h.SH configure.com handy.h !> win32/config.bc win32/config.gc win32/config.vc !> win32/config.vc64 win32/config_H.bc win32/config_H.gc !> win32/config_H.vc win32/config_H.vc64 ____________________________________________________________________________ [ 28233] By: nicholas on 2006/05/19 15:01:03 Log: Integrate: [ 28007] Subject: two [PATCH]es: dVAR for malloc.c, and a signed/unsigned char* From: Jarkko Hietaniemi Date: Fri, 28 Apr 2006 21:04:59 +0300 Message-ID: <4452594B.4040609@gmail.com> [the signed/unsigned part] [ 28042] Subject: [PATCH] signed vs unsigned char * nit From: Jarkko Hietaniemi Date: Sat, 29 Apr 2006 20:59:29 +0300 (EEST) Message-Id: <200604291759.k3THxTji032009@kosh.hut.fi> Branch: maint-5.8/perl !> pp.c ____________________________________________________________________________ [ 28220] By: nicholas on 2006/05/17 21:36:24 Log: Update Changes Branch: maint-5.8/perl ! Changes patchlevel.h ____________________________________________________________________________ [ 28213] By: nicholas on 2006/05/16 22:44:31 Log: Integrate: [ 28034] Fix two errors found by Coverity. Branch: maint-5.8/perl !> pp_sys.c ____________________________________________________________________________ [ 28212] By: nicholas on 2006/05/16 21:23:48 Log: Integrate: The perlio.c part of: [ 25915] Subject: [PATCH] Etta James presents: More consting From: Andy Lester Date: Sun, 30 Oct 2005 23:44:13 -0600 Message-ID: <20051031054413.GA10767@petdance.com> [ 25948] Subject: [PATCH] perlio.c consting From: Andy Lester Date: Oct 31, 2005 6:54 PM Message-ID: <20051031175411.GA18480@petdance.com> Branch: maint-5.8/perl !> op.c perlio.c ____________________________________________________________________________ [ 28211] By: nicholas on 2006/05/16 17:29:06 Log: Integrate: [ 27484] The "Change installation prefix, if necessary." code needs to use the ~ expanded prefix and installprefix in the substitution. [ 27485] The existing order of the checks for whether man1dir and man3dir were ' ' were obviously buggy, because they set installman[13]dir only for the next if block to set it again. Reordering removes that daftness. [ 27488] Refactor most of the cut & paste for setting prefix/prefixexp variables and the "Change installation prefix, if necessary" into a new "function" setprefixvar (which I assume will become Setprefixvar.U) How many platforms still have Bourne shells that can't do shell functions? Someone's going to say Irix? [ 27489] Split break out the installprefix code from setprefixvar, which allows the remaining 8 places to call the common prefix conversion code. [ 27490] First stab at rather shaky Configure support for relocatable @INC. Seems to be installing rather more that it's supposed to be. [ 27491] Somewhat less shaky relocatable @INC support. You can install things to the current location of the perl tree, move it, install more things, move it again, lather rinse repeat. Configure with -Duserelocatableinc [ 27513] Make the original versions of relocated paths available from %Config::Config with the prefix "raw_". Not sure if "raw_" is the best choice. [ 27603] make configpm only update Config.pm, Config_heavy.pl if changed. Should stop spurious rebuilding of extensions. Also add documentation and copyright to top of file. [ 27703] Regeneration after backports of #27447, #27484, #27485, #27488, #27489, #27490, #27491, and #27632 [ 27758] Regeneration of Glossary and the underlying tools. [ 28188] Only ask about MAD for 5.9.0 and greater. Otherwise silently select 'n'. [ 28196] Simplify the implementation in ExtUtils::Packlist::read. [ 28198] Encode run-time relocation of file names in packlist with a relocate_as attribute. With this, unmodified instmodsh correctly verifies installed modules even after relocation. [ 28202] Take advantage of the bug that binexp ignores the Configure installprefix, by using initialinstalllocation rather than binexp to determine the #! line for installed scripts. This way they are at least correct initially for a relocatable @INC perl, rather than starting as .../../bin/perl [ 28210] Revert change 27513 [Make the original versions of relocated paths available from %Config::Config with the prefix "raw_". Not sure if "raw_" is the best choice.] Branch: maint-5.8/perl !> Configure Porting/Glossary config_h.SH configpm installperl !> lib/ExtUtils/Packlist.pm ____________________________________________________________________________ [ 28208] By: nicholas on 2006/05/16 16:09:59 Log: Integrate: [ 27016] Subject: Re: [PATCH] Re: replacing "inuse" Win files (was Re: Help with a Cwd.pm build error) From: demerphq Date: Tue, 31 Jan 2006 09:55:25 +0100 Message-ID: <9b18b3110601310055h7aeb9aa1gdc2d63cc56d97768@mail.gmail.com> [ 27587] Get version numbers in line with what's in the CPAN EU::MM 6.30_01 tarball. (ExtUtils::Install has local changes, so bump its version number.) [ 27983] Upgrade to ExtUtils-Install-1.39 [ 28101] Upgrade to ExtUtils-Install-1.40 Branch: maint-5.8/perl !> lib/ExtUtils/Install.pm lib/ExtUtils/Installed.pm !> lib/ExtUtils/Packlist.pm lib/ExtUtils/t/Install.t !> lib/ExtUtils/t/Installed.t lib/ExtUtils/t/Packlist.t ____________________________________________________________________________ [ 28207] By: nicholas on 2006/05/16 13:57:39 Log: Integrate: [ 27407] Add a Configure question for mad (Misc Attribute Decoration - Larry's perl5 to perl[56] convertor), which if set defines PERL_MAD in config.h [ 27410] Backport #27407 Add a Configure question for mad (Misc Attribute Decoration - Larry's perl5 to perl[56] convertor), which if set defines PERL_MAD in config.h [ 27632] Introduce d_sitearch and d_inc_version_list config.sh variables. These allow me to set sitearch and sitelib to be the same without ending up with duplicate entries in @INC. Basically the same way the old d_archlib variable is used. [ 27634] Populate pre-canned config.sh files with d_archlib and d_inc_version_list. Ref change 27632. [ 27652] Glossary part for #27447 (madlyh, madlyobj, madlysrc) Branch: maint-5.8/perl !> Configure Cross/config.sh-arm-linux NetWare/config.wc !> Porting/Glossary config_h.SH configure.com epoc/config.sh !> perl.c plan9/config_sh.sample uconfig.sh win32/config.bc !> win32/config.gc win32/config.vc win32/config.vc64 !> wince/config.ce ____________________________________________________________________________ [ 28204] By: nicholas on 2006/05/16 12:57:39 Log: Integrate: [ 26757] Subject: Re: relaxing TEST restrictions (was: Re: 5.9.3 approaches) From: Yitzchak Scott-Thoennes Date: Mon, 9 Jan 2006 06:38:25 -0800 Message-ID: <20060109143825.GB4132@efn.org> [ 26959] Revert change 24461, now that change 26757 allows TEST to accept extra output from module tests [ 28192] TEST needs to binmode the file handle that is reading the TAP, else it can get killed if TEST's environment has made it default all opens to UTF-8 but the TAP is not UTF-8. Branch: maint-5.8/perl !> cygwin/perlld.in t/TEST ____________________________________________________________________________ [ 28203] By: nicholas on 2006/05/16 12:08:39 Log: Integrate: [ 26048] 3 calls to strlen() we can avoid. [ 26049] Eliminate duplicate strlen()s by explicitly calling it. Note two places for future audit with FIXME. Convert a SvPVX_const to SvPVX_const_nolen, where the length is unused. [ 26050] Eliminate some unnecessary strlen()s [ 26051] Moving the 4 writes of PL_no_mem followed by exit(1) to a static function makes the object code slightly smaller. [ 26052] It helps to actually *remove* the now-superfluous strlen(). (An addendum to 26050.) [ 26059] Remove superfluous strlen() from pp_require(). [ 26061] Simplify Perl_allocmy slightly, and cope better with the name is "" case. [ 26180] Eliminate two more strlen()s, by using the return value of sprintf(). [ 26188] Subject: [PATCH] ff_chop.patch redux From: Andy Lester Date: Mon, 21 Nov 2005 21:16:14 -0600 Message-ID: <20051122031614.GB19424@petdance.com> [ 26231] Subject: [PATCH] Remove redundant SvUTF8_on() calls From: Gisle Aas Date: 30 Nov 2005 09:28:54 -0800 Message-ID: Branch: maint-5.8/perl !> doio.c op.c perl.c perlio.c pp.c pp_ctl.c sv.c toke.c util.c ____________________________________________________________________________ [ 28201] By: nicholas on 2006/05/15 16:44:06 Log: Integrate: [ 24081] [with modification - the extra argument to incpush was supposed to be being used] Branch: maint-5.8/perl ! perl.c ____________________________________________________________________________ [ 28199] By: nicholas on 2006/05/15 16:20:18 Log: Integrate: [ 23667] Move the setting of $^X ahead of the initialisation of @INC [ 23668] Move the definition of the S_procself_val() function before the point where it's used [ 23674] Relocatable @INC entries for Unix. (With appropriate fixups in Config.pm to complete the illusion) Currently can only be enabled with hackery to config.sh TODO - proper Configure support, and support for otherlibdirs in Config.pm [ 23675] Oops. You can't add new entries to config_h.SH without making some sort of corresponding defined/undefined value in config.sh somehow. [ 23676] Bother. It helps if you also propagate the new entry to config.sh [ 23677] Subject: [PATCH] Re: Smoke [5.9.2] 23675 FAIL(m) openvms E8.2 (Alpha/4 cpu) From: "Craig A. Berry" Message-ID: <41CB6BD1.8030207@mac.com> Date: Thu, 23 Dec 2004 19:07:29 -0600 [ 23679] Don't do the ... expansion on archlib, sitelib etc, given that everyone should already be using archlibext etc Add the extra special case code for otherlibdirs [ 23696] Rebuilt after backport start for 23674, 23675, and 23676 (and the configpm part of) [ 24271] Subject: [PATCH] Symbian port of Perl From: Date: Mon, 18 Apr 2005 13:18:30 +0300 Message-ID: Branch: maint-5.8/perl !> Configure Porting/Glossary configpm configure.com embed.fnc !> embed.h perl.c proto.h ____________________________________________________________________________ [ 28191] By: nicholas on 2006/05/14 11:01:39 Log: Integrate: [ 27902] Upgrade to Test::Harness 2.57_05 [ 27925] Subject: Re: TODO tests and test::harness From: demerphq Date: Thu, 20 Apr 2006 22:36:08 +0200 Message-ID: <9b18b3110604201336k5a974f28h732a2819853b995@mail.gmail.com> [ 27940] Assimilate Test::Harness 2.57_06 [ 28190] Upgrade to Test::Harness 2.58. Branch: maint-5.8/perl +> lib/Test/Harness/Util.pm !> MANIFEST lib/Test/Harness.pm lib/Test/Harness/Assert.pm !> lib/Test/Harness/Changes lib/Test/Harness/Point.pm !> lib/Test/Harness/Straps.pm lib/Test/Harness/TAP.pod !> lib/Test/Harness/bin/prove lib/Test/Harness/t/inc_taint.t !> lib/Test/Harness/t/prove-globbing.t !> lib/Test/Harness/t/prove-switches.t lib/Test/Harness/t/strap.t !> lib/Test/Harness/t/test-harness.t lib/Test/Harness/t/version.t ____________________________________________________________________________ [ 28184] By: nicholas on 2006/05/12 21:35:03 Log: Casts to cope with non-const function prototypes in maint. Branch: maint-5.8/perl ! doio.c pp_ctl.c pp_sys.c ____________________________________________________________________________ [ 28182] By: nicholas on 2006/05/12 18:29:48 Log: Integrate: [ 27176] Ensure that public I, N and P flags are off when SvTAINT is called on something that already has taint magic. [ 27219] Convert utftaint.t to test.pl (from Test.pm) and provide it with tainted() from taint.t to remove the dependency on Scalar::Util. (So it will now work with miniperl) [ 27221] Turn of tainting locally inside swash_init(), as the internal implementation of the regexp engine has nothing to do with user data, so when and where it happens to call require is agnostic of the current expression's taintedness. [ 27222] save_re_context() and even errsv_save = newSVsv(ERRSV); can turn PL_tainted back on, so defer turning if off as late as possible. This gets lib/locale.t working once more. Branch: maint-5.8/perl !> sv.c t/op/utftaint.t utf8.c ____________________________________________________________________________ [ 28170] By: nicholas on 2006/05/11 21:22:41 Log: Integrate: [ 25804] Subject: [PATCH] blead 25801: Symbian batch of today From: Date: Wed, 19 Oct 2005 16:19:23 +0300 Message-ID: [ 26030] Subject: [PATCH] Symbian update blead@26025 From: Date: Mon, 7 Nov 2005 13:32:15 +0200 Message-ID: Branch: maint-5.8/perl !> doio.c perl.c perl.h pp_sys.c ____________________________________________________________________________ [ 28169] By: nicholas on 2006/05/11 20:23:59 Log: Integrate: [ 27329] Subject: Re: [perl #38619] Bug in lc and uc (interaction between UTF-8, substr, and lc/uc) From: SADAHIRO Tomoyuki Date: Sat, 25 Feb 2006 18:16:45 +0900 Message-Id: <20060225180934.FCC3.BQW10602@nifty.com> [ 28006] Fix bug 34297 (length of overloaded UTF-8 strings) [ 28011] lc plus an 8 bit locale could mangle UTF-8 values returned by overloaded stringification. [ 28012] uc plus an 8 bit locale could get confused by UTF-8 values returned by overloaded stringification. [ 28013] lcfirst/ucfist plus an 8 bit locale could mangle UTF-8 values returned by overloaded stringification. [ 28014] There were more ways to break uc/lc/ucfirst/lcfirst even without use 'locale'; so test for them too. Correct the skip count. [ 28016] print couldn't correctly handle surprises from UTF-8 overloading. [ 28019] syswrite couldn't correctly handle surprises from UTF-8 overloading. As part of fixing this, syswrite now tries to take advantage of the UTF-8 cache logic for lengths and offsets on regular scalars. [ 28022] index and rindex couldn't correctly handle surprises from UTF-8 overloading. [ 28027] Dave noticed that change 28022 was wrong - should read is_index rather than just index. [ 28028] We had no tests that covered the grow paths in pp_uc and pp_lc. [ 28029] do_vop() couldn't correctly handle surprises from UTF-8 overloading. [ 28030] One part of pp_pack couldn't correctly handle surprises from UTF-8 overloading. Branch: maint-5.8/perl + t/uni/overload.t !> MANIFEST doio.c doop.c pp.c pp_pack.c pp_sys.c t/op/lc.t ____________________________________________________________________________ [ 28167] By: nicholas on 2006/05/11 16:08:47 Log: Die die die die die! And stay dead, please. Branch: maint-5.8/perl - ext/Encode/META.yml ____________________________________________________________________________ [ 28165] By: nicholas on 2006/05/11 16:01:23 Log: Integrate: [ 25821] Mark more static Encode data structures as const. [ 25823] use more 'const' in the Encode data structures. [ 26081] Pod nit in Encode.pm, found by Marc Lehmann in RT #36949. [ 26295] Don't build manpages for Encode and Unicode::Normalize (when compiling perl manpages are built afterwards, not by MakeMaker like when installing modules.) This fix a bootstrap problem (now that Pod::Man uses POSIX.) [ 26863] Upgrade to Encode 2.14 [ 26922] Avoid warning with MS Visual C compiler. Encode.xs(443) : warning C4761: integral size mismatch in argument; conversion supplied [ 27721] There should be only one META.yml file [ 27824] Coverity notes that resource are leaked in the unfinished and never called function _utf8_to_bytes. [ 27982] Upgrade to Encode-2.15 [ 28098] Upgrade to Encode 2.16 [ 28141] Ressurect change 27824, which plugs a resource leak in uncalled code. [ 28152] Upgrade to Encode-2.17 Branch: maint-5.8/perl +> ext/Encode/t/from_to.t - ext/Devel/PPPort/META.yml !> (integrate 70 files) ____________________________________________________________________________ [ 28164] By: nicholas on 2006/05/11 13:34:17 Log: Integrate: [ 26352] The logic to use SVt_NV or SVt_PVNV is in sv_upgrade, so no need to dupliate it in sv_2nv. Also, the final sv_upgrade is never called. [ 26353] Simplify the SvGMAGIC code in sv_2nv, removing duplicated checks to warn for uninitialized values. [ 26354] Simplify the SvGMAGIC code in sv_[ipu]v_flags. [ 26355] Duplicate the AMAGIC temporary variable consting from sv_2iv_flags into sv_2uv_flags, sv_2nv and sv_2pv_flags. [ 26385] The THINKFIRST check after the GMAGICAL check in sv_2iv_flags and sv_2uv_flags should have been else if. [ 26386] The IV/UV choice can be made inside uiv_2buf rather than sv_2pv_flags. [ 26390] Pull the definition of olderrno in sv_2pv_flags into the block where it's used. Branch: maint-5.8/perl !> sv.c ____________________________________________________________________________ [ 28163] By: nicholas on 2006/05/11 13:08:38 Log: Integrate: [ 26002] You can't there from here. (Dead code in sv_2pv_flags) [ 26003] Use the return value of sprintf in sv_2pv_flags. Remove dead code. [ 26267] Untease the regexp stringification from the reference naming in sv_2pv_flags. (Lots of re-indentation, little real change) [ 26268] As they are now the same, can fold the entire switch statement for reference type in sv_2pv_flags into a call to sv_reftype. [ 26269] Clone the brief return logic and thereby remove a goto. [ 26270] Move the scope of origsv inwards, and rename variables to avoid the need to re-assign to sv. [ 26271] Make all the return statements closer to the final calcuation of the returned value in the reference stringification code. Pull variables into tigher scope which also allows some to become const. [ 26272] sv_2pv_flags shouldn't return a constant string "NULLREF" - it was the only constant non-"" string returned. [ 26273] Eliminate all the gotos in sv_2pv_flags, by moving the tokensave* code into the GMAGIC block. [ 26274] Eliminate an unneeded local variable. [ 26275] Don't use Copy for 1 and 2 character string constants. [ 26290] Pull the regexp stringification code out into S_stringify_regexp, to make sv_2pv_flags clearer. [ 26345] Inline asIV and asUV, as each is only used once, and it distracts from the patterns of repeated code in sv_2iv_flags, sv_2uv_flags, sv_2nv and sv_2pv_flags. Add a comment noting the return path from the end of the SvGMAGICAL case to all 4. [ 26346] Update embed.fnc and headers after change 26345 (spotted by Rafael) Branch: maint-5.8/perl ! sv.c !> embed.fnc embed.h proto.h ____________________________________________________________________________ [ 28162] By: nicholas on 2006/05/11 12:41:36 Log: Integrate: [ 28131] Subject: [PATCH] ext/IPC/SysV/t/ipcsysv.t using test.pl From: David Landgren Message-ID: <445B694B.8060901@landgren.net> Date: Fri, 05 May 2006 17:03:39 +0200 Subject: Re: [PATCH] ext/IPC/SysV/t/sem.t using test.pl From: David Landgren Message-ID: <445B75EF.3000100@landgren.net> Date: Fri, 05 May 2006 17:57:35 +0200 [ 28138] Subject: Re: [PATCH] ext/IPC/SysV/t/ipcsysv.t using test.pl From: David Landgren Date: Tue, 09 May 2006 13:03:22 +0200 Message-ID: <446076FA.6010409@landgren.net> Branch: maint-5.8/perl !> ext/IPC/SysV/t/ipcsysv.t ext/IPC/SysV/t/sem.t ____________________________________________________________________________ [ 28161] By: nicholas on 2006/05/11 12:16:42 Log: Integrate: [ 27220] Assume that if runperl is called under tainting, that the caller really really wanted to run perl, so brute force untaint everything. [ 27345] Move all the de-tainting logic for runperl into test.pl. [ 27364] Taint handling for runperl: - better taint detection (switch -T in command) - $ENV{PATH} stripping of writeable directories on unix/linux [ 27550] Also strip $ENV{PATH} dirs writable by world. The stripping was introduced in change 27364 to improve taint handling. [ 27551] Avoid "uninitialized value" warnings during 'make test'. My system is set up with a few PATH entries I'm not allowed to access, so stat() fails for these. [ 28037] Subject: Re: [PATCH] provide diag() and don't use ++ in test.pl From: David Landgren Date: Mon, 01 May 2006 15:50:03 +0200 Message-ID: <4456120B.3000302@landgren.net> Plus some additional comment (also from David) about WHY not to use ++ in test.pl [ 28059] Change 28037 broke t/op/each.t when running with TEST. Branch: maint-5.8/perl !> t/test.pl ____________________________________________________________________________ [ 28160] By: nicholas on 2006/05/11 12:14:58 Log: Integrate: [ 25615] Micro-optimisation: use PL_globalstash instead of looking up CORE::GLOBAL::require by name [ 25848] Code in wait4pid was calling hv_delete with the hash iterator currently on that entry. On aggregate this does more work, beacuse the next call to hv_iterinit() would spot the flag, and have to call the delete routine, while in the meantime any new entries can't re-use that memory. [ 25854] Subject: Re: [PATCH] minor speed-up for pp_push() From: Tassilo von Parseval Date: Tue, 25 Oct 2005 08:11:42 +0200 Message-id: <20051025061141.GA2256@ethan> Branch: maint-5.8/perl !> op.c pp.c util.c ____________________________________________________________________________ [ 28159] By: nicholas on 2006/05/11 11:55:15 Log: Integrate: [ 24958] Indent some of the #ifdef/#define/#endif to make the nesting clearer. Simplify SvPV_free slightly. [just the second part] Branch: maint-5.8/perl !> sv.h ____________________________________________________________________________ [ 28158] By: nicholas on 2006/05/11 10:36:47 Log: Integrate: [ 25832] Add my_sprintf, which is usually just a macro for sprintf, for those places where we want to use the return value of sprintf. This allows a wrapper to be used for platforms where the C library isn't ANSI- conformant. [ 25841] Replace all the strlen()s related to PL_pidstatus with the return value of my_sprintf [ 25862] Use the return value from sprintf and avoid a call to strlen [ 26001] Use the return value from sprintf(). [ 26006] s/printf/my_printf/ because we're using the return value. My mistake spotted by Gisle. Branch: maint-5.8/perl !> embed.fnc embed.h global.sym makedef.pl perl.h proto.h !> regcomp.c sv.c taint.c universal.c util.c ____________________________________________________________________________ [ 28157] By: nicholas on 2006/05/11 10:03:53 Log: Integrate: [ 23763] Add TODO tests for UTF8 encoded soft references [ 23766] Stage 1 of utf8 support for soft references. Change gv_fetchpv to take a UTF8 flag, as gv_fetchpvn_flags Add gv_fetchsv to look up a GV by SV rather than a char * pointer Provide a backwards compatability gv_fetchpv Migrate from gv_fetchpv to gv_fetchsv where the caller was grabbing the pointer from an SV All tests still pass. [ 23770] Pull the am-I-utf8-or-not logic into one place (S_newSV_maybe_utf8) as I think that it will be needed for utf8 soft references. [ 26415] gv_fetchpv, gv_fetchpvn and gv_fetchsv take a bitmask of flags, rather than a simple boolean, so passing FALSE or TRUE is bogus. [ 26434] The lref argument of sv_2cv is actually passed onwards to gv_fetchsv, so it is a bitmap of flag bits rather than simple TRUE/FALSE. [ 27028] Make Perl_gv_fetchpvn_flags actually heed the passed in length. This means that \0 bytes in symbolic references now work. [ 27044] Oops. 1 NUL termination assumption remains in Perl_gv_fetchpvn_flags [ 27046] doubleplusoops. Apart from the documented NUL termination assumption. Now removed. [ 27049] And another assumption. Branch: maint-5.8/perl ! pp_sys.c !> doio.c embed.fnc embed.h global.sym gv.c gv.h mg.c op.c perl.c !> perl.h pp.c pp_hot.c proto.h regcomp.c sv.c sv.h t/op/ref.t !> toke.c universal.c ____________________________________________________________________________ [ 28154] By: nicholas on 2006/05/10 18:31:34 Log: Integrate: [ 25746] The new REQUIRE token introduced by change 25599 must be nonassoc, just like the UNIOP token it's patterned after. (While we're at it, allow to use bison 2.1 to regenerate the parser files.) plus fix change 28153 - rename the two parameter Perl_dofile() to Perl_dofile2(), and make Perl_dofile() a wrapper to it just in case anything is linking to it. Branch: maint-5.8/perl ! embed.fnc embed.h op.c perly.c perly.y perly_c.diff proto.h ! vms/perly_c.vms !> perly.h ____________________________________________________________________________ [ 28153] By: nicholas on 2006/05/10 17:44:54 Log: Integrate: [ 25599] CORE::require was always parsed as require(). That's because require() isn't overridable at tokenizer-level like other overridable built-ins, but is handled by the optree builder. So, find a way to pass the information that require() was written as CORE::require() to Perl_ck_require. This is done by adding a new token type REQUIRE and by adding OPf_SPECIAL to OP_REQUIRE when it's saw as CORE::require in the program text. This fixes bug [perl #37274] The "CORE" in CORE::require is ignored. [ 25600] Document the new meaning of OPf_SPECIAL on OP_REQUIRE [ 25611] Test for OPpCONST_NOVER only on OP_CONST ops. Plus a regression test by Schwern. [ 25616] Fix the overriding of CORE::do, just like change 25599 was fixing the overriding of CORE::require Branch: maint-5.8/perl ! perly.c perly_c.diff vms/perly_c.vms vms/perly_h.vms !> embed.fnc embed.h op.c op.h perly.h perly.y pp_ctl.c proto.h !> t/op/override.t toke.c ____________________________________________________________________________ [ 28151] By: nicholas on 2006/05/10 14:36:06 Log: Integrate: [ 25344] Add the "no 6" / "no v6" syntax. [ 25346] Oops, fix threaded build [ 28149] Convert use.t to an inlined is/isnt/like implementation, to give better diagnostics. [ 28150] no 5.9.4; should fail in version 5.9.4. Improve the diagnostics and test names in t/comp/use.t Branch: maint-5.8/perl ! ext/B/t/concise-xs.t t/comp/use.t !> embed.fnc embed.h op.c op.h pp_ctl.c proto.h toke.c ____________________________________________________________________________ [ 28147] By: nicholas on 2006/05/09 22:15:36 Log: Integrate: [ 25166] Subject: Re: [perl #27028] /$/ not honouring /m in some cases From: Rick Delaney Date: Jul 14, 2005 4:10 PM Message-ID: <20050714141059.GF19090@localhost.localdomain> Branch: maint-5.8/perl +> t/op/regexp_qr_embed.t !> MANIFEST t/op/regexp.t ____________________________________________________________________________ [ 28130] By: nicholas on 2006/05/08 20:44:45 Log: Integrate: [ 27802] Replace explict "/dev/null" with BIT_BUCKET, which varies by platform. [ 27804] In the description of require, clarify the file handle return and why C must be qualified. [ 27811] Add a test for source filters returned from code references in @INC. [ 27812] Add more tests for the builtin source filter implementation, and fix two bugs - it would loop infinitely if data were already in the read buffer, and it would process those data twice. [ 27813] Tests for having multiple pipes open simultaneously. (See comment added in change 3771 about this apparently being buggy) [ 27814] If you return void from a filter subroutine, the caller's status is used. If not, your last statement is the return value, and if it keeps being +ve you'll never hit EOF. The probably should be a sanity check for this in toke.c, but I'm not sure how. [ 27815] Eliminate the filter_child_proc hack from pp_require. [ 27816] If the downstream caller wants block mode, and we're in line mode, then don't return more bytes than they asked for. Hold bytes over until next time if necessary. [ 27817] $foo .= $bar doesn't warn if $foo is undefined, so simplify code. [ 27819] Augment the user filter caching code so that if the user filter returns multiple lines, only one line at a time is returned. Rename the variable len to status, as it is the status value. [ 27820] Fix a bug - the introduction of the caching code means that we should initialise the SV upstream after it. assert that filter_sub is always non-null now. [ 27821] I believe that it makes little sense for the simple user filters provided by coderefs to be run in block mode. Hence always read from the upstream filter in line mode, and always call the user's filter with 1 line. [ 27827] Teach Perl_do_sv_dump how to cope with IOs that have bogus things in GV slots. (Which source filters are prone to do to them). [ 27849] A scalar reference returned from a coderef in @INC is treated as the initial "content" of the file. When it is exhausted input is taken from a real file handle, or a generator sub, if either exists. [ 27879] Document all the return values that code references in @INC can return. [ 27880] Coderefs in @INC are done and documented. [ 27881] Fix an error, spotted by Tim Bunce. [ 28080] Subject: -Dmad breaks a few tests From: Sébastien Aperghis-Tramoni Date: Tue, 2 May 2006 01:37:03 +0200 Message-Id: <64F8ECEA-D96B-11DA-AA80-000502F3279F@free.fr> (part of, with tweaks) Branch: maint-5.8/perl +> t/op/incfilter.t ! sv.h !> MANIFEST dump.c pod/perlfunc.pod pod/perltodo.pod pp_ctl.c !> t/op/inccode.t ____________________________________________________________________________ [ 28129] By: nicholas on 2006/05/08 20:15:44 Log: Integrate: [ 27334] Subject: [PATCH] Speedups and shrinkages of SvREFCNT_inc From: andy@petdance.com (Andy Lester) Date: Fri, 24 Feb 2006 14:54:35 -0600 Message-ID: <20060224205434.GA17867@petdance.com> [ 27348] Silence a GCC warning (assignment from incompatible pointer type) [ 27701] Subject: [PATCH] SvREFCNT_inc tweaks From: andy@petdance.com (Andy Lester) Date: Sat, 1 Apr 2006 02:01:30 -0600 Message-ID: <20060401080130.GA19372@petdance.com> Branch: maint-5.8/perl ! gv.c op.c perl.c perlio.c !> av.c cop.h doio.c doop.c hv.c mg.c pad.c perl.h !> pod/perlapi.pod pod/perlintern.pod pp.c pp_ctl.c pp_hot.c !> pp_sort.c pp_sys.c regcomp.c scope.c sv.c sv.h toke.c ____________________________________________________________________________ [ 28128] By: nicholas on 2006/05/08 19:22:03 Log: Integrate: [ 26380] s/Nullav/NULL/g [ 26381] s/Nullhv/NULL/g; # Although I see that Robin is proposing to add some :-) [ 27054] Subject: Re: [PATCH] s/Null(av|ch)/NULL/g From: Steven Schubiger Date: Thu, 2 Feb 2006 10:38:49 +0100 Message-ID: <20060202093849.GD12591@accognoscere.homeunix.org> [ 27065] Subject: Re: [PATCH] s/Null(gv|hv|sv)/NULL/g From: Steven Schubiger Message-ID: <20060203152449.GI12591@accognoscere.homeunix.org> Date: Fri, 3 Feb 2006 16:24:49 +0100 [ 27067] Oops. When changing warnings.pl one should re-run it. Well spotted Rafael. [ 27078] warnings.h does some evil(*) pointer arithmetic on (SV *)0, so a simple replace of Nullsv with NULL turns out to be a bad idea. * Technically undefined behaviour, I believe. [ 27238] More NullXXX macro removal from Andy Lester Branch: maint-5.8/perl ! op.c !> (integrate 46 files) ____________________________________________________________________________ [ 28124] By: nicholas on 2006/05/08 18:08:05 Log: Tidy up warnings about implicitly removing qualifiers from pointers. Branch: maint-5.8/perl ! sv.c ____________________________________________________________________________ [ 28123] By: nicholas on 2006/05/08 17:43:39 Log: Integrate: [ 27564] (Hopefully) make VMS happy. [ 27607] Silence 3 (spurious) gcc warnings about "maybe used uninitialized". [Well, the 2 of them applicable to maint] Branch: maint-5.8/perl !> sv.c ____________________________________________________________________________ [ 28121] By: nicholas on 2006/05/08 15:53:23 Log: Integrate: [ 27525] Add a new per-interpeter variable PL_utf8cache, which will be used to control the UTF-8 offset caching code. Make this visible as ${^UTF8CACHE} [ 27536] Properly set/reset the UTF-8 length cache in regcomp.c. [ 27537] Remove a duplicated conditional test by putting everything for the not readonly case inside one block. [ 27538] Make Perl_sv_len_utf8 honour the new ${^UTF8CACHE}. If PERL_UTF8_CACHE_ASSERT is defined, default ${^UTF8CACHE} to -1 (assertion mode). Need a way to turn this on with existing command line flags. [ 27543] Remove a redundant test on mg->mg_ptr. (It had to be non-NULL to get into this block). [ 27552] The two loops in Perl_utf8_distance can be merged. [ 27553] Perl_utf8_distance can be implemented as a call to Perl_utf8_length. [ 27557] Abstract the string walking functionality from Perl_sv_pos_b2u into static functions, and make it respect PL_utf8cache. [ 27558] In Perl_sv_pos_b2u, if we know the character offset of the end, we can take advantage of this to count backwards if close. Remove some now unnecessary casts. [ 27559] Extend change 27558 - if we know the character offset of the end, but don't have a cache for intermediate offsets, we can still use the end offset. [ 27570] The simplest way to start on Perl_sv_pos_u2b seems to be to take out all of the caching code. [ 27572] Create S_sv_pos_u2b_cached, which abstracts away utf-8 to byte conversions, hiding the complexity of any cache lookup. [ 27573] Add S_sv_pos_u2b_midway for when we know the utf-8/bytes offsets on both sides of the desired offset, so could count either way. Use cached utf-8 length in S_sv_pos_u2b_cached. Check the results if PL_utf8cache < 0. [ 27574] Take advantage of uoffset >= uoffset0, to simplify some logic. [ 27575] Call S_utf8_mg_pos_cache_update() from S_sv_pos_u2b_cached(). S_utf8_mg_pos_cache_update() can't use S_sv_pos_b2u_forwards() in the cache assertion code as it triggers warnings about surrogates. [ 27578] Use the cache in S_sv_pos_u2b_cached when the sought offset is after the first cache entry. [ 27579] Use the cache in S_sv_pos_u2b_cached() when the desired offset is between the start of the string and the cached location. Remove unneccesary else from Perl_sv_pos_b2u(). [ 27580] Add all new static functions to embed.fnc Remove old unused static functions. Remove references to cache[2] and cache[3]. [ 27582] Pass the (byte) length of the entire string into utf8_mg_pos_cache_update() Start to use the cache to store two pairs of byte/utf-8 pairs. Add the first third of the cache update code. [ 27583] Complete the other 2 arms of the cache update code. [ 27584] Use both cache entries for Perl_sv_pos_b2u(). [ 27585] Make S_sv_pos_u2b_cached use both cache pairs. [ 27586] Move orphaned documentation back to its functions Document the new static functions, and how the cache works. [ 27588] Rejig the location of the cache assertion code in Perl_sv_pos_b2u(). [ 27589] Document the values for ${^UTF8CACHE} [ 27875] Turn on UTF8 cache assertions with -Ca Branch: maint-5.8/perl !> embed.fnc embed.h embedvar.h gv.c intrpvar.h locale.c mg.c !> perl.c perl.h perlapi.h pod/perlrun.pod pod/perltodo.pod !> pod/perlvar.pod proto.h regcomp.c sv.c utf8.c util.c ____________________________________________________________________________ [ 28120] By: nicholas on 2006/05/08 15:31:45 Log: Subject: Re: Smoke [5.8.8] 28115 FAIL(M) MSWin32 WinXP/.Net SP2 (x86/2 cpu) From: Steve Hay Message-ID: <445F2CED.3010808@uk.radan.com> Date: Mon, 08 May 2006 12:35:09 +0100 Branch: maint-5.8/perl ! makedef.pl ____________________________________________________________________________ [ 28119] By: nicholas on 2006/05/08 13:45:02 Log: Integrate: [ 27728] Save the regexp engine state as 1 block on the save stack. [ 27739] Teach ss_dup about SAVEt_RE_STATE. (As correctly as it ever was before) [ 27930] Given that use re 'debug' can turn on DEBUGGING even on a non-DEBUGGING build, we cannot assume that PL_reg_starttry isn't needed. So copy it in Perl_ss_dup. Branch: maint-5.8/perl ! regexp.h scope.c !> regcomp.c scope.h sv.c ____________________________________________________________________________ [ 28117] By: nicholas on 2006/05/07 23:30:16 Log: Hateful gcc. gc, I feel. Branch: maint-5.8/perl ! mg.c ____________________________________________________________________________ [ 28115] By: nicholas on 2006/05/07 00:03:20 Log: Integrate: [ 25193] Exit test script (if we're going to) *before* declaring a plan [ 25200] Subject: [PATCH] support POSIX SA_SIGINFO From: Jarkko Hietaniemi Date: Wed, 20 Jul 2005 14:40:54 +0300 Message-ID: <42DE3846.6050606@gmail.com> [ 25203] Fix definition of Perl_sighandler following change 25200 [ 25207] Subject: [PATCH] Re: [PATCH] support POSIX SA_SIGINFO From: Jarkko Hietaniemi Date: Thu, 21 Jul 2005 18:33:22 +0300 Message-ID: <42DFC042.1090201@gmail.com> [ 25223] use POSIX might not throw an error. Check for useful values also [ 25230] Some signals are more real than others Also added comment about why the tests are there [ 25240] Subject: RE: [PATCH] Re: [PATCH] support POSIX SA_SIGINFO From: "Green, Paul" Date: Mon, 25 Jul 2005 12:39:32 -0400 Message-ID: [ 25248] SIGRTMIN & SIGRTMAX on Cygwin both 32. sig_count = 33, but POSIX::SigRt is not available on use [ 25586] Subject: [PATCH] quiet a few warnings From: Yitzchak Scott-Thoennes Date: Wed, 21 Sep 2005 18:45:56 -0700 Message-ID: <20050922014555.GA2852@efn.org> [ 25975] Test for the right signal should use the constant for that signal. [ 25976] Missing va_end(args) from SA_SIGINFO related code. [ 25977] Change #25976 seemed to cause test failures in ext/B/t/f_sort.t. What was that line about Jenga again? [ 25980] Reverting the reversion. va_end() goes with va_start(). [ 27440] croak in POSIX::sigaction() when passed a negative signal instead dumping core. [ 27441] Need to update test count from change #27440. [ 28039] Disable SIGRTMAX and SIGRTMIN in POSIX if Configure's probes found that they aren't viable. (Certain glibcs are "interesting") Branch: maint-5.8/perl ! embed.fnc ext/POSIX/Makefile.PL mg.c perl.c perl.h perlvars.h !> cygwin/cygwin.c embed.h ext/POSIX/POSIX.pod ext/POSIX/POSIX.xs !> ext/POSIX/t/sigaction.t global.sym iperlsys.h pp_sys.c proto.h !> util.c ____________________________________________________________________________ [ 28114] By: nicholas on 2006/05/06 10:35:42 Log: Integrate: [ 26449] Break out the item normalisation code into a method normalise_items. [ 26451] Break out the code to generate #ifdef/#endif into new methods macro_to_ifdef and macro_to_endif Add an args hashref to normalise_items; provide a disable_utf8_duplication argument to disable the utf8 duplication code. [ 26453] Make ExtUtils::Constant generate the inlineable proxy constant subs. So far just enough to make Fcntl work. [ 26455] Use inlineable proxy constant subs for POSIX. There may be trouble ahead, as it seems that not all POSIX "constants" are. I wonder if too many systems are going to have too many variations to make this viable. [ 26463] Use inlineable proxy constant subs for Socket [ 26464] SIGRTMAX, SIGRTMIN, LDBL_MAX, LDBL_MIN, LDBL_EPSILON and HUGE_VAL aren't constants on (at least some) Linux. And more evil than that, HUGE_VAL likes to use the C preprocessor in its value, necessitating use of a temporary. [ 26465] Need to be more careful with the symbol table manipulation - if there is already an entry in place, back off and generate a real constant subroutine. (fixes lib/Net/hostent.t failure) Fix a bug with how hv_store was being called. [ 26472] Add support for PV to ExtUtils::Constant::ProxySubs, and enable its use in Sys::Syslog [ 26475] Provide support for types PVN and UNDEF in ExtUtils::Constant::ProxySubs [ 26488] The values for proxy constant subroutines need to be read only. [ 26490] Add C_FH and XS_FH arguments to ExtUtils::Constant::WriteConstants, to allow the caller to pass in file handles. Use this in Contant.t with tied file handles to capture the output, rather than calling the lower level routines directly. [ 26490] Add C_FH and XS_FH arguments to ExtUtils::Constant::WriteConstants, to allow the caller to pass in file handles. Use this in Contant.t with tied file handles to capture the output, rather than calling the lower level routines directly. [ 26491] Add a couple of newlines for tidier regen output, consistent with the code generated by WriteConstants() [ 26495] Add method macro_from_item to encapsulate the entire logic for getting the correct macro direct from the item hash. [ 26496] Support type UNDEF, and correct a bug in the macro generation by using macro_from_item. [ 26497] Remove the hard wired test numbers in the generated test script for simple tests. Use this script twice, firstly for testing the old style autoloaded constants, then for testing the ProxySubs. Make testing the dogfood/regeneration code optional, as the ProxySubs output doesn't contain it. [ 26502] As what we're doing is generating subroutines, which are methods, we need to invalidate cached methods. For constant.pm we need to create &Internals::inc_sub_generation [ 26506] Pass in the full arguments to ExtUtils::Constant::ProxySubs Prototype all the missing constants, so that parsing doesn't depend on the phase of the moon. (Well, the system headers) Add an option for making missing constants generate errors at read time (which can be during constant folding). This isn't the default. [ 26507] s/%/%%/ in the package name so that it can be used safely in the sprintf format string (and hence not need runtime %s interpolation) [ 26679] Negating an unsigned value generates warnings for some compilers, so cast it first. [ 26932] constant_add_symbol needs to be static, else static builds can fail at link time. (*How* many meanings is "static" overloaded with?) [ 27242] Need to clear CvXSUBANY() too to turn a constant sub into a prototype. [ 27720] Fix test portability on non-ASCII platforms Subject: Re: lib/ExtUtils/t/Constant.t problem on z/OS From: SADAHIRO Tomoyuki Date: Wed, 05 Apr 2006 21:13:02 +0900 Message-Id: <20060405210426.F99D.BQW10602@nifty.com> [ 27962] Subject: Re: [PATCH] cleanup 212 warnings emitted by gcc-4.2 From: Marcus Holland-Moritz Date: Mon, 24 Apr 2006 23:20:38 +0200 Message-ID: <20060424232038.7550f9b6@r2d2> [ 28038] Allow a macro of "0" to mean "#if 0" for convenient elimination of constants. Branch: maint-5.8/perl +> lib/ExtUtils/Constant/ProxySubs.pm !> MANIFEST lib/ExtUtils/Constant.pm !> lib/ExtUtils/Constant/Base.pm lib/ExtUtils/Constant/XS.pm !> lib/ExtUtils/t/Constant.t ____________________________________________________________________________ [ 28113] By: nicholas on 2006/05/05 21:14:29 Log: Integrate: [ 25185] Subject: [PATCH] allow POSIX SIGRTMIN...SIGRTMAX signals (and plug a core dump) From: Jarkko Hietaniemi Date: Tue, 19 Jul 2005 12:06:00 +0300 Message-ID: <42DCC278.2010009@gmail.com> [ 25188] XS constants fail when their value is called for, not at import. This might be construed as a bug. Fixing is probably a greater evil. [ 25189] C inside Foo.pm considered a "bad thing" (Certainly where there is AUTOLOADing and XS) Hopefully this fixes the POSIX breakage. [ 25190] C inside Foo.pm considered a "bad thing" (Certainly where there is AUTOLOADing and XS) Hopefully *this* fixes the POSIX breakage. [ 25191] Subject: [PATCH] move NSIG logic From: Jarkko Hietaniemi Date: Wed, 20 Jul 2005 09:57:44 +0300 Message-ID: <42DDF5E8.1060100@gmail.com> [ 25199] Silence warnings about NSIG redefinition on Win32 (Change 25191 doesn't work out unless dosish.h includes too, which it currently only does for DJGPP) Branch: maint-5.8/perl !> Configure dosish.h ext/POSIX/Makefile.PL ext/POSIX/POSIX.pm !> ext/POSIX/POSIX.pod ext/POSIX/POSIX.xs ext/POSIX/t/sigaction.t !> handy.h perl.h ____________________________________________________________________________ [ 28112] By: nicholas on 2006/05/05 20:21:42 Log: Integrate: [ 25688] Subject: Re: $^CHILD_ERROR_NATIVE issues (with attachment) From: Gisle Aas Date: 04 Oct 2005 02:18:27 -0700 Message-ID: [ 25772] Subject: [patch@25763] Fix VMS error/exit handling, update kill function From: "John E. Malmberg" Message-id: <4351F393.8030809@qsl.net> Date: Sun, 16 Oct 2005 02:30:43 -0400 [ 25810] Subject: [patch@25809]restore documented exit behavior From: "John E. Malmberg" Date: Thu, 20 Oct 2005 18:21:20 -0400 Message-id: <43581860.3020108@qsl.net> [ 25839] Subject: VMS exit handling still broken, need some help. From: "John E. Malmberg" Date: Mon, 24 Oct 2005 01:34:41 -0400 Message-ID: <435C7271.8070403@qsl.net> [ 25849] Back out change #25839, and apply : Subject: [patch@25838]Hopefully the last VMS exit/error fixes needed. From: "John E. Malmberg" Date: Tue, 25 Oct 2005 11:09:37 -0400 Message-ID: <435E4AB1.1000106@qsl.net> [ 25851] Subject: Re: [patch@25838]Hopefully the last VMS exit/error fixes needed. From: "John E. Malmberg" Date: Tue, 25 Oct 2005 13:36:20 -0400 Message-ID: <435E6D14.7000104@qsl.net> Branch: maint-5.8/perl ! scope.h !> cop.h doio.c mg.c perl.c perl.h pp_sys.c t/op/exec.t !> t/run/exit.t vms/perlvms.pod vms/vms.c vms/vmsish.h !> win32/perlhost.h wince/perlhost.h ____________________________________________________________________________ [ 28111] By: nicholas on 2006/05/05 19:41:27 Log: Integrate: [ 24613] Subject: [PATCH] VMS build update for blead From: "Craig A. Berry" Date: Sat, 28 May 2005 23:18:46 -0500 Message-Id: <429942A6.10601@mac.com> [ 24731] Remove unneeded #ifdef VMS since VMS is not defined here anyway (spotted by Steve_p) [ 25256] Various patches by John E. Malmberg to fix data corruption issues on VMS. Back out change 25218 and the parts of change 25217 that affect pp_ctl.c and pp_sys.c. [ 25760] Subject: patch for blead - Perlvms.pod update From: "John E. Malmberg" Date: Fri, 14 Oct 2005 21:15:26 -0400 Message-id: <4350582E.4070209@qsl.net> [ 25761] Subject: [patch@blead] perlport.pod From: "John E. Malmberg" Date: Fri, 14 Oct 2005 23:48:20 -0400 Message-id: <43507C04.8050003@qsl.net> Branch: maint-5.8/perl ! vms/vms.c !> doio.c mg.c pod/perlport.pod pp_ctl.c pp_sys.c util.c !> vms/perlvms.pod vms/vmsish.h ____________________________________________________________________________ [ 28110] By: nicholas on 2006/05/05 18:59:13 Log: Integrate: [ 25306] Subject: [patch@25305] fixes to const fixes + Case Preserved start From: "John E. Malmberg" Date: Thu, 18 Aug 2005 21:18:27 -0400 Message-ID: <43053363.8090809@qsl.net> Branch: maint-5.8/perl !> vms/vms.c ____________________________________________________________________________ [ 28109] By: nicholas on 2006/05/05 17:26:10 Log: Integrate: [ 25257] Various patches by John E. Malmberg to fix data corruption issues on VMS. Back out change 25218 and the parts of change 25217 that affect pp_ctl.c and pp_sys.c. [ 25265] VMS debugging help, by John E. Malmberg [ 25280] Subject: patch@25279 VMS error handling and const fixes From: "John E. Malmberg" Date: Wed, 10 Aug 2005 00:37:13 -0400 Message-ID: <42F98479.6030207@qsl.net> [Bowlderised to keep the non-static functions in embed.fnc unchanged] Branch: maint-5.8/perl ! pp_sys.c !> doio.c embed.fnc mg.c perl.h pp_ctl.c proto.h !> vms/descrip_mms.template vms/vms.c vms/vmsish.h ____________________________________________________________________________ [ 28107] By: nicholas on 2006/05/05 16:20:20 Log: Integrate: [ 24501] Subject: Well defined $? and introduction of ${^CHILD_ERROR_NATIVE} [PATCH] From: Gisle Aas Date: 18 May 2005 08:35:47 -0700 Message-ID: [ 24503] Adjust VMS test count [ 24506] Fix the new exit() tests, by Gisle Aas : don't let the shell interfere with the process return value [ 24507] Gisle noted an unused variable [ 24606] Skip signals and core dump tests on Win32 See http://www.xray.mpe.mpg.de/mailing-lists/perl5-porters/2005-05/msg00749.html [ 24691] POD fixes [ 25687] Subject: Clone PL_statusvalue_posix [PATCH] From: Gisle Aas Date: 03 Oct 2005 09:37:57 -0700 Message-ID: [ 25688] Subject: Re: $^CHILD_ERROR_NATIVE issues (with attachment) From: Gisle Aas Date: 04 Oct 2005 02:18:27 -0700 Message-ID: Branch: maint-5.8/perl !> doio.c embedvar.h gv.c intrpvar.h mg.c perl.c perl.h perlapi.h !> pod/perlfunc.pod pod/perlport.pod pod/perlvar.pod sv.c !> t/op/exec.t t/run/exit.t ____________________________________________________________________________ [ 28002] By: nicholas on 2006/04/28 14:12:06 Log: Integrate: [ 26569] add svt_local slot to magic vtable, and fix local $shared [ 26735] document svt_copy, svt_dup and svt_local vtable slots Branch: maint-5.8/perl !> ext/threads/shared/shared.xs mg.c mg.h pod/perlguts.pod ____________________________________________________________________________ [ 27996] By: nicholas on 2006/04/28 13:21:30 Log: Integrate: [ 24264] Subject: [PATCH] Fix MGVTBL structures in perl.h From: Andy Lester Date: Wed, 20 Apr 2005 15:09:24 -0500 Message-Id: <20050420200923.GA3017@petdance.com> The MGVTBL structures contain 7 function pointers, but only 5 were getting initialized. Plus, there were repeated lists between declaration and definition; add a macro to declare or define all the global MGVTBLs as appropriate. [ 24268] Subject: patch to [PATCH] Fix MGVTBL structures in perl.h (preprocessor From: David Dyck Date: Wed, 20 Apr 2005 16:35:55 -0700 (PDT) Message-Id: fix nits in change #24264: some preprocessors don't like #ifdefs within a macro's args; also removed duplicate semicolons from macro definitions. [ 25060] Subject: Re: [PATCH] Re: [PATCH] Re: a blead warning from tru64 From: Jarkko Hietaniemi Date: Mon, 04 Jul 2005 00:19:28 +0300 Message-Id: <42C85660.50807@gmail.com> Branch: maint-5.8/perl !> perl.h ____________________________________________________________________________ [ 27985] By: nicholas on 2006/04/27 19:14:04 Log: I didn't want ExtUtils::CBuilder. Only ExtUtils::ParseXS, and that's merely as a side effect of xsubpp now being a thin wrapper around it. Branch: maint-5.8/perl - lib/ExtUtils/CBuilder.pm lib/ExtUtils/CBuilder/Base.pm - lib/ExtUtils/CBuilder/Platform/Unix.pm - lib/ExtUtils/CBuilder/Platform/VMS.pm - lib/ExtUtils/CBuilder/Platform/Windows.pm - lib/ExtUtils/CBuilder/Platform/aix.pm - lib/ExtUtils/CBuilder/Platform/cygwin.pm - lib/ExtUtils/CBuilder/Platform/darwin.pm - lib/ExtUtils/CBuilder/Platform/dec_osf.pm - lib/ExtUtils/CBuilder/Platform/os2.pm - lib/ExtUtils/CBuilder/t/01-basic.t - lib/ExtUtils/CBuilder/t/02-link.t - lib/ExtUtils/ParseXS/t/XSTest.pm - lib/ExtUtils/ParseXS/t/XSTest.xs - lib/ExtUtils/ParseXS/t/basic.t ! MANIFEST ____________________________________________________________________________ [ 27984] By: nicholas on 2006/04/27 18:09:09 Log: Integrate: [ 24820] Make some variables global, to avoid some "will not stay shared" warnings at compile time [ 25222] Make CBuilder and ParseXS clean up their temp test files [ 25225] Ken pointed out that CBuilder's cleanup is too agrressive The Windows compile() and link() methods bizarrely include their own output files in the list of files to be cleaned up. Now that they actually are being cleaned up, this isn't ideal ;-) Stop compile() and link() from cleaning up their own output, and reinstate the explicit deletion of those output files in the test scripts. [ 25533] Upgrade to ExtUtils::ParseXS 2.12 [ 25690] Upgrade to ExtUtils-ParseXS-2.13 [ 25722] Upgraded to ExtUtils-ParseXS-2.14. [ 25764] Upgrade to ExtUtils-ParseXS-2.15. Branch: maint-5.8/perl !> lib/ExtUtils/ParseXS.pm lib/ExtUtils/ParseXS/t/basic.t ____________________________________________________________________________ [ 27981] By: nicholas on 2006/04/27 17:48:33 Log: Integrate: [ 24500] Subject: Re: [PATCH] ExtUtils-{ParseXS,CBuilder} into bleadperl (was: Re: [Module::Build] ANNOUNCE: Module::Build 0.2610 -> CPAN) From: Yitzchak Scott-Thoennes Date: Wed, 18 May 2005 07:13:40 -0700 Message-ID: <20050518141131.GA2704@efn.org> [ 24504] tabs, not spaces. Bad editor. No cookie. Does the MANIFEST file support comments, sufficient to add an editor block? [ 24537] Subject: [PATCH] ExtUtils::CBuilder Tru64 support From: Jarkko Hietaniemi Date: Sat, 21 May 2005 23:20:21 +0300 Message-Id: <428F9805.70108@gmail.com> [ 24558] Make distclean should remove the leftover test files of ExtUtils::ParseXS [ 24568] Subject: [PATCH] xsubpp, ParseXS: allow other suffix beyond .c From: jarkko.hietaniemi@nokia.com Date: Tue, 24 May 2005 12:48:38 +0300 Message-Id: Branch: maint-5.8/perl +> lib/ExtUtils/CBuilder.pm lib/ExtUtils/CBuilder/Base.pm +> lib/ExtUtils/CBuilder/Platform/Unix.pm +> lib/ExtUtils/CBuilder/Platform/VMS.pm +> lib/ExtUtils/CBuilder/Platform/Windows.pm +> lib/ExtUtils/CBuilder/Platform/aix.pm +> lib/ExtUtils/CBuilder/Platform/cygwin.pm +> lib/ExtUtils/CBuilder/Platform/darwin.pm +> lib/ExtUtils/CBuilder/Platform/dec_osf.pm +> lib/ExtUtils/CBuilder/Platform/os2.pm +> lib/ExtUtils/CBuilder/t/01-basic.t +> lib/ExtUtils/CBuilder/t/02-link.t lib/ExtUtils/ParseXS.pm +> lib/ExtUtils/ParseXS/t/XSTest.pm +> lib/ExtUtils/ParseXS/t/XSTest.xs +> lib/ExtUtils/ParseXS/t/basic.t !> MANIFEST Makefile.SH lib/ExtUtils/xsubpp ____________________________________________________________________________ [ 27980] By: nicholas on 2006/04/27 15:30:40 Log: Fix up a couple more errors in 2 static functions' prototypes, and add NN to 3 other functions. Branch: maint-5.8/perl ! embed.fnc proto.h ____________________________________________________________________________ [ 27979] By: nicholas on 2006/04/27 14:38:12 Log: Pesky stowaways. Branch: maint-5.8/perl - perly.tab ____________________________________________________________________________ [ 27978] By: nicholas on 2006/04/27 14:37:43 Log: Integrate: [ 24412] while (my $x ...) { ...; redo } shouldn't undef $x. In the presence of 'my' in the conditional of a while(), until(), or for(;;) loop, add an extra scope to the body so that redo doesn't undef the lexical Branch: maint-5.8/perl +> perly.tab ! embed.fnc op.c perly.c perly.y perly_c.diff t/op/loopctl.t ! vms/perly_c.vms !> embed.h perly.h pp_ctl.c proto.h ____________________________________________________________________________ [ 27977] By: nicholas on 2006/04/27 14:19:31 Log: Fix up my mistakes with prototypes made while merging pack changes. Prototypes for non-static functions are now restored to 5.8.8. Branch: maint-5.8/perl ! mathoms.c pp_pack.c proto.h ____________________________________________________________________________ [ 27975] By: nicholas on 2006/04/27 12:22:13 Log: Integrate: [ 24023] Subject: [PATCH] pad_push numero three-o From: Andy Lester Date: Thu, 10 Mar 2005 15:17:28 -0600 Message-ID: <20050310211728.GC32386@petdance.com> Branch: maint-5.8/perl !> embed.fnc pad.c pod/perlintern.pod ____________________________________________________________________________ [ 27974] By: nicholas on 2006/04/27 11:41:22 Log: Integrate: [ 25779] Make some casts explicit to keep VC++ 7 happy Specifically, this silences the warnings from the following smoke: http://www.nntp.perl.org/group/perl.daily-build.reports/32258 (Hmm. Looks like you need to decode the base64 yourself before you can read that.) [ 26032] Subject: [PATCH] Missing static declarations From: Gisle Aas Date: 05 Nov 2005 05:08:03 -0800 Message-ID: [ 26130] The warning "Field too wide in 'u' format in pack" was emitted even when pack warnings were not enabled [ 26242] Subject: [PATCH] Signs, signs, everywhere a sign From: Andy Lester Date: Thu, 1 Dec 2005 14:24:51 -0600 Message-ID: <20051201202451.GI22713@petdance.com> [ 26549] Subject: [PATCH] op/pack.t failure on Win64 From: Jan Dubois Date: Dec 30, 2005 9:55 PM Message-ID: <017a01c60d83$5dc73110$2217a8c0@candy> [ 27956] Perl_unpack_str() is a mathom! Branch: maint-5.8/perl !> genpacksizetables.pl mathoms.c pp_pack.c ____________________________________________________________________________ [ 27973] By: nicholas on 2006/04/26 23:15:54 Log: Integrate: [ 24965] Subject: [PATCH] Pre-YAPC consting fun From: Andy Lester Date: Jun 23, 2005 9:04 PM Message-ID: <20050623190423.GA13835@petdance.com> [ 25129] make the expensive ckWARN() be called as late as possible reorganise if (ckWARN(FOO) && should_not_happen_condition) to if (should_not_happen_condition && ckWARN(FOO)) Branch: maint-5.8/perl !> pp_pack.c ____________________________________________________________________________ [ 27972] By: nicholas on 2006/04/26 22:56:05 Log: Integrate: [ 24328] Subject: Re: [PATCH] pp_pack.c redux, to work w/Jarkko's recent patches From: Andy Lester Date: Mon, 25 Apr 2005 17:56:51 -0500 Message-ID: <20050425225651.GB17271@petdance.com> [ 24350] Subject: (in blead) a missing const in a prototype From: Date: Wed, 27 Apr 2005 17:35:27 +0300 Message-ID: [ 24772] Subject: [PATCH] pp_pack.c cleanup From: Andy Lester Date: Wed, 8 Jun 2005 10:23:53 -0500 Message-ID: <20050608152353.GA6253@petdance.com> [ 24778] More SvPV consting Branch: maint-5.8/perl !> embed.fnc embed.h perl.h pod/perlapi.pod pp_pack.c proto.h ____________________________________________________________________________ [ 27971] By: nicholas on 2006/04/26 22:23:19 Log: Integrate: [ 24061] Casts needed to make VMS happy. (It gets very upset at assignments between char * and unsigned char *) [ 24126] Subject: [PATCH] pp_pack.c warnings on VMS From: "Craig A. Berry" Date: Thu, 31 Mar 2005 14:51:05 -0600 Message-ID: <424C62B9.2030601@mac.com> [ 24251] Avoid generating thousands of UTF-8 warnings when testing under a UTF-8 locale [ 24279] Subject: [PATCH] pp_pack.c warning quieting From: Andy Lester Date: Thu, 21 Apr 2005 11:33:13 -0500 Message-ID: <20050421163313.GA20930@petdance.com> Branch: maint-5.8/perl !> pp_pack.c t/op/pack.t ____________________________________________________________________________ [ 27970] By: nicholas on 2006/04/26 22:04:33 Log: Integrate: [ 24100] Subject: Re: PATCH: byte count feature request for unpack From: perl5-porters@ton.iguana.be (Ton Hospel) Date: Sun, 27 Mar 2005 18:32:11 +0000 (UTC) Message-Id: (rework of a patch from Arne Ahrend ) [Code is merged in, but the new features are disabled in maint] Branch: maint-5.8/perl ! pp_pack.c t/op/pack.t !> pod/perlfunc.pod ____________________________________________________________________________ [ 27969] By: nicholas on 2006/04/26 20:43:36 Log: Integrate: [ 24052] Subject: pack / for general types From: perl5-porters@ton.iguana.be (Ton Hospel) Date: Sat, 19 Mar 2005 22:00:45 +0000 (UTC) Message-Id: Allow "len/format" to work for any format type, not just strings. [ 24060] Subject: Re: unpack A strip patch From: Ton Hospel Date: Mon, 21 Mar 2005 21:31:37 +0000 (UTC) Message-Id: Branch: maint-5.8/perl ! pod/perlfunc.pod pp_pack.c t/op/pack.t !> pod/perldiag.pod ____________________________________________________________________________ [ 27967] By: nicholas on 2006/04/26 18:48:16 Log: Integrate: [ 24031] Subject: Re: Encoding neutral unpack From: perl5-porters@ton.iguana.be (Ton Hospel) Date: Sat, 12 Mar 2005 18:54:29 +0000 (UTC) Message-ID: A new test. [ 24038] Subject: pp_pack.c, simplifying genpacksizetables From: perl5-porters@ton.iguana.be (Ton Hospel) Date: Sun, 13 Mar 2005 15:07:41 +0000 (UTC) Message-Id: Branch: maint-5.8/perl !> genpacksizetables.pl pp_pack.c t/op/pack.t ____________________________________________________________________________ [ 27966] By: nicholas on 2006/04/26 18:21:44 Log: Integrate: [ 24011] Fix memory corruption when growing pack utf8 buffer [ 24012] Further pack optimisations by Ton Hospel [ 24030] Indentation patch by Ton Hospel for pp_pack Branch: maint-5.8/perl ! pp_pack.c ____________________________________________________________________________ [ 27965] By: nicholas on 2006/04/26 17:51:25 Log: Integrate: [ 24010] Subject: Encoding neutral unpack From: perl5-porters@ton.iguana.be (Ton Hospel) Date: Sun, 6 Mar 2005 18:29:38 +0000 (UTC) Message-Id: (Bowlderised for maint, bug compatibility braindeath in preference to Ton's elegant correctness) Branch: maint-5.8/perl ! pod/perldiag.pod pod/perlfunc.pod pod/perlunicode.pod ! pod/perluniintro.pod pp_pack.c t/op/pack.t !> embed.fnc embed.h genpacksizetables.pl lib/charnames.t perl.h !> proto.h ____________________________________________________________________________ [ 27964] By: nicholas on 2006/04/26 14:32:50 Log: Integrate: [ 23988] Subject: [PATCH] signed/unsigned help for pp_pack.c From: "Craig A. Berry" Date: Fri, 18 Feb 2005 09:42:52 -0600 Message-ID: <42160CFC.3080304@mac.com> Branch: maint-5.8/perl !> pp_pack.c ____________________________________________________________________________ [ 27963] By: nicholas on 2006/04/26 14:16:33 Log: Integrate: [ 23966] Subject: Re: encoding neutral unpack From: perl5-porters@ton.iguana.be (Ton Hospel) Date: Sat, 5 Feb 2005 01:34:44 +0000 (UTC) Message-ID: tweaked to remove the 'not supported on this platform' error messages Tweaked for maint quite a lot: * Don't provide W * Keep the old meanings of U0 and C0 * Only change the UTF-8 behaviour on numeric templates Branch: maint-5.8/perl ! genpacksizetables.pl pp_pack.c t/op/pack.t ____________________________________________________________________________ [ 27929] By: nicholas on 2006/04/20 22:13:54 Log: Silence a compiler warning by (of all things) removing a cast. Branch: maint-5.8/perl ! pp_hot.c ____________________________________________________________________________ [ 27928] By: nicholas on 2006/04/20 21:48:52 Log: Integrate: [ 24942] handle magic in local correctly the local SV now gets a copy of any container magic, and no value magic; in the past the whole magic chain was either shared or moved [ 24943] remove taint hack now that local $tainted no longer copies taint magic [ 25081] change 24943 broke restoration of localized taint values [ 25094] more taint fallout from change 24943 Branch: maint-5.8/perl ! mg.c pod/perlguts.pod !> embed.fnc embed.h pod/perlintern.pod proto.h scope.c sv.c !> t/op/local.t t/op/taint.t ____________________________________________________________________________ [ 27926] By: nicholas on 2006/04/20 21:03:43 Log: Integrate: [ 23925] Attempt to fix problems with new exception handling macros. Branch: maint-5.8/perl +> ext/XS/APItest/exception.c !> MANIFEST XSUB.h ext/XS/APItest/APItest.xs !> ext/XS/APItest/MANIFEST ext/XS/APItest/Makefile.PL !> pod/perlguts.pod ____________________________________________________________________________ [ 27924] By: nicholas on 2006/04/20 20:46:31 Log: Integrate: [ 23911] Add simple exception handling macros for XS writers. [ 23913] Regenerate perlapi.pod. [ 27560] Test croak(NULL) Branch: maint-5.8/perl +> ext/XS/APItest/t/exception.t !> MANIFEST XSUB.h ext/XS/APItest/APItest.pm !> ext/XS/APItest/APItest.xs pod/perlapi.pod pod/perlguts.pod ____________________________________________________________________________ [ 27921] By: nicholas on 2006/04/20 20:17:41 Log: Integrate: [ 27660] Rewrite t/op/getppid.t to remove the assumption that init has PID of 1. Branch: maint-5.8/perl !> t/op/getppid.t ____________________________________________________________________________ [ 27920] By: nicholas on 2006/04/20 19:58:11 Log: Integrate: [ 23474] Skip failing thread/wait tests on HP-UX 10.20 These will never ever be fixed, since 10.20 is obsolete perl583delta describes this. The quote is included in wait.t Branch: maint-5.8/perl !> ext/threads/shared/t/wait.t ____________________________________________________________________________ [ 27916] By: nicholas on 2006/04/20 17:07:07 Log: Integrate: [ 27704] Two modules in ext/ have dependencies that are post-miniperl. Encode this information in the Makefile for now. Branch: maint-5.8/perl !> Makefile.SH ____________________________________________________________________________ [ 27915] By: nicholas on 2006/04/20 16:06:43 Log: Integrate: [ 23018] Assert that we aren't leaking memory. Branch: maint-5.8/perl !> sv.c ____________________________________________________________________________ [ 27914] By: nicholas on 2006/04/20 15:43:10 Log: Integrate: [ 20479] Fix bug #23141 : localization of readonly magic scalars now produces an error "Modification of a read-only value attempted", instead of silently failing. [ 21323] When localising a magic value, propagate the readonly flag only if this scalar has \0 magic or has magic without a 'set' method. (follows change #20479 for bug #23141.) [ 22578] Good catch from Dave Mitchell; plus a null pointer protection. Branch: maint-5.8/perl !> scope.c t/op/local.t ____________________________________________________________________________ [ 27743] By: nicholas on 2006/04/08 18:07:27 Log: Make the fold_constants exception capture compatible with -DPERL_FLEXIBLE_EXCEPTIONS Branch: maint-5.8/perl ! embed.fnc embed.h op.c proto.h ____________________________________________________________________________ [ 27742] By: nicholas on 2006/04/08 17:28:28 Log: Integrate: [ 27615] If constant folding fails, don't fold constants, rather than reporting the error at compile time. This allows illegal constant expressions in dead code to be ignored. [ 27616] It helps to actually add the files you add to MANIFEST. (Thanks Dave for spotting my mistake). Branch: maint-5.8/perl +> t/comp/fold.t !> MANIFEST op.c pod/perldiag.pod pod/perltodo.pod ____________________________________________________________________________ [ 27624] By: nicholas on 2006/03/28 19:35:56 Log: Integrate: [ 22819] Add new dUNDERBAR and UNDERBAR macros, to help XS writers to access the $_ variable, even when there is a lexical one in scope. (but use the ppport.h compatibility definitions of the two, and change the documentation to make it clear that lexical $_ is 5.9.2 and later) Branch: maint-5.8/perl ! XSUB.h pod/perlapi.pod ____________________________________________________________________________ [ 27623] By: nicholas on 2006/03/28 16:42:11 Log: Integrate: [ 21729] Subject: [PATCH] Re: Storable Error From: Nicholas Clark Date: Sat, 9 Aug 2003 00:35:36 +0100 Message-ID: <20030809003535.C20130@plum.flirble.org> [ 22300] The new -Dq option didn't actually work! [ 27622] Only compile Perl_hv_assert with DEBUGGING. Branch: maint-5.8/perl !> embed.fnc hv.c makedef.pl perl.c perl.h pod/perlrun.pod !> proto.h ____________________________________________________________________________ [ 27614] By: nicholas on 2006/03/27 16:31:27 Log: Actually make PERL_FLEXIBLE_EXCEPTIONS compile (although it still fails 3 tests - t/op/eval.t, t/op/sort.t and ext/XS/APItest/t/call.t) Branch: maint-5.8/perl ! perl.c pp_ctl.c ____________________________________________________________________________ [ 27606] By: nicholas on 2006/03/25 17:59:37 Log: Integrate: [ 26673] Remove code duplicated a few lines above by change 19695. [ 26760] Avoid possible dereference of NULL in the initialization of PL_origalen. This can only happen when perlparse is called with no argv. Don't try to update PL_origargv unless PL_origalen is at least 2. [ 26779] Make setting 'PL_origalen = 1' before perl_parse() disable argv[0] munging when $0 is assigned to. Branch: maint-5.8/perl !> mg.c perl.c pod/perlembed.pod toke.c ____________________________________________________________________________ [ 27605] By: nicholas on 2006/03/25 17:42:10 Log: Integrate: [ 21538] Fix off-by-one in $0 set magic. (From Autrijus.) The whole PL_origalen thing in perl.c looks very hokey. [ 21546] Will the real off by one please stand up? [ 21664] Orphaned submission from #21546. [ 21833] Subject: [PATCH] setting $0 still broken From: Jan Dubois Date: Thu, 27 Nov 2003 21:09:51 -0800 Message-ID: Branch: maint-5.8/perl !> mg.c perl.c ____________________________________________________________________________ [ 27604] By: nicholas on 2006/03/25 15:15:28 Log: Integrate: [ 23471] Subject: [PATCH blead] [perl #3038] Re: $qr = qr/^a$/m; $x =~ $qr; fails From: Rick Delaney Date: Sun, 31 Oct 2004 22:40:40 -0500 Message-ID: <20041101034040.GC1232@biff.bort.ca> Branch: maint-5.8/perl +> t/op/regexp_qr.t ! regexec.c !> MANIFEST pp.c pp_hot.c t/op/regexp.t ____________________________________________________________________________ [ 27512] By: nicholas on 2006/03/15 20:21:52 Log: When bless() changes the overloading state of the passed in reference, and there are other references to this object, then brute force search for the other references, and perform the same change on them. Fixes bug 34925. Integrate the regression tests from: [ 27506] Moving the overloading flag from the reference to the referant allows (re)?blessing of overloaded objects to work correctly. [ 27508] Test for reblessing objects with weak references. [ 27510] D'oh! 27508 wasn't quite testing what I wanted it to test. This does. Branch: maint-5.8/perl ! embed.fnc embed.h lib/overload.t proto.h sv.c ____________________________________________________________________________ [ 27509] By: nicholas on 2006/03/15 18:14:26 Log: Integrate: The overload.t tests from 23357 (Turn regcomp into a list op) [ 27501] Partially convert overload.t to Test::More [ 27502] Further conversion of overload.t [ 27504] Further conversion of overload.t [ 27505] Finish converting overload.t to Test::More Branch: maint-5.8/perl ! lib/overload.t ____________________________________________________________________________ [ 27426] By: nicholas on 2006/03/08 21:24:41 Log: Avoid writing over the input string in the case 'F' in moreswitches. Branch: maint-5.8/perl ! perl.c ____________________________________________________________________________ [ 27395] By: nicholas on 2006/03/06 21:28:10 Log: sv_2iv and sv_2uv can go in mathoms.c now. Branch: maint-5.8/perl ! mathoms.c sv.c ____________________________________________________________________________ [ 27394] By: nicholas on 2006/03/06 21:11:18 Log: Integrate: [ 26265] The extra return is actually a duplicated code path, and can go. [ 26285] The early return for SvIOKp(sv) in sv_2[iu]v_flags is actually code duplication. [ 26286] Move vast swathes of common code from sv_2iv_flags and sv_2uv_flags into S_sv_2iuv_common. [ 26289] 3 instances of SvIsUV_on(sv); can be replaced with one. Branch: maint-5.8/perl ! sv.c ____________________________________________________________________________ [ 27392] By: nicholas on 2006/03/06 20:40:15 Log: Integrate: [ 25322] Subject: [PATCH blead] Use SvGETMAGIC more often From: Rick Delaney Date: Mon, 22 Aug 2005 13:00:33 -0400 Message-ID: <20050822170033.GH7674@localhost.localdomain> [ 26185] Subject: [PATCH] concat interacts badly with magic From: Robin Houston Date: Mon, 21 Nov 2005 17:03:39 +0000 Message-ID: <20051121170339.GA8214@rpc142.cs.man.ac.uk> [ 26190] Revert change #26185, which could have some unwanted side-effects. (lib/warnings.t had failures due to this patch.) Moreover Dave Mitchell correctly pointed out that [perl #37722] wasn't really a bug. Branch: maint-5.8/perl !> doio.c pp.c pp_ctl.c pp_hot.c sv.c t/op/tie.t universal.c !> xsutils.c ____________________________________________________________________________ [ 27391] By: nicholas on 2006/03/06 18:13:50 Log: Integrate: [ 22074] Some of the bitwise manipulation PP functions weren't checking whether their arguments were magical before using them. (bug [#24816]). [ 22163] Subject: Re: [perl #24816] Magic vars seem unsure if they are purely numeric From: Yitzchak Scott-Thoennes Date: Thu, 15 Jan 2004 14:10:37 -0800 Message-Id: <20040115221037.GA2392@efn.org> Add sv_2iv_flags() to allow magic to be optionally processed. Branch: maint-5.8/perl ! t/op/bop.t !> doop.c embed.fnc embed.h global.sym pod/perlapi.pod pp.c !> proto.h sv.c sv.h ____________________________________________________________________________ [ 27318] By: nicholas on 2006/02/24 17:07:53 Log: Integrate: [ 26990] Avoid using space to create 2 functions just to force mathoms to be linked. Instead, force the link via the perl interpreter structure via a function pointer that gets replaced at run time anyway. Branch: maint-5.8/perl !> embed.fnc embed.h intrpvar.h mathoms.c perl.c proto.h ____________________________________________________________________________ [ 27317] By: nicholas on 2006/02/24 16:36:54 Log: Integrate: [ 26007] Merge pp_andassign, pp_orassign, and pp_dorassign into pp_and, pp_or, and pp_defined, respectively. [ 26008] Typos in opcode.pl this time. [ 26015] Allow passing of the full enum name into the tryAMAGICbin family of macros, to avoid needing C pre-processor string concatenation within the lowest level expansion. [ 26039] Merge pp_slt, pp_sgt and pp_sge into pp_sle. (The most commonly used of the 4. Not that any are that popular.) [ 26589] The two never used NOTYET switch ops can go. [ 27071] Silence a warning about "statement not reached" from the Sun C compiler. It was accurate - the conditional code used to give 2 return statements. [ 27082] I believe that mathoms have been completely TODOne. [ 27106] Move Perl_save_long, Perl_save_I16, Perl_save_I8, Perl_save_iv, Perl_save_nogv, Perl_save_list, Perl_save_destructor to mathoms.c Perl_save_svref can't actually move because it calls a static function. [ 27113] Optimise index so that if the big string is ISO-8859-1 but the little string is UTF-8, it tries to downgrade the little string, rather than upgrade the big string. For half-meg big strings this is a fourfold speed gain. [ 27116] Fix bug 38454 (rindex corrects for $[ on bytes rather than UTF-8) [ 27117] Some refactoring to converge pp_index and pp_rindex [ 27118] Merge pp_index and pp_rindex - we have another mathom. [ 27119] pp_hex can be implemented trivially by pp_oct, making pp_hex a mathom. [ 27121] pp_pop can be implemented by pp_shift. [ 27122] Create new macros AMG_CALLun_var, AMG_CALLun_var and tryAMAGICun_var which don't do the pre-processor string manipulation internally. [ 27123] Avoid gcc warning about possibly uninitialised variables. [ 27124] All the trancendental unary operators can be merged into PP_sin (cos, exp, log, sqrt) Branch: maint-5.8/perl ! mathoms.c pp.c scope.c utf8.c !> opcode.h opcode.pl pod/perltodo.pod pp.h pp_ctl.c t/op/index.t ____________________________________________________________________________ [ 27316] By: nicholas on 2006/02/24 15:19:21 Log: Integrate: [ 26007] Merge pp_andassign, pp_orassign, and pp_dorassign into pp_and, pp_or, and pp_defined, respectively. [ 26008] Typos in opcode.pl this time. [ 26035] The rarely used lcfirst and ucfirst share almost all their code. Merge the two as pp_ucfirst. [ 26057] Don't declare a function inside another function [ 26058] Patch from Gisle to add NO_MATHOMS to the Compile-time options. [ 26077] Define PERL_DEFAULT_DO_EXEC3_IMPLEMENTATION earlier in perl.h (that is, before embed.h, that uses this symbol, is included) [ 26099] Subject: [PATCH] mathoms prototypes aren't prototypes From: Andy Lester Date: Fri, 11 Nov 2005 17:20:30 -0600 Message-ID: <20051111232030.GA882@petdance.com> Branch: maint-5.8/perl ! mathoms.c pp.c !> opcode.h opcode.pl perl.c perl.h pp_ctl.c pp_hot.c ____________________________________________________________________________ [ 27315] By: nicholas on 2006/02/24 14:55:44 Log: Integrate: [ 25983] The PERL_EFF_ACCESS_[RWX]_OK macros are all defined as triples, so can be refactored to a single platform dependent PERL_EFF_ACCESS macro with the other 3 implemented in terms of it. [ 25984] PERL_EFF_ACCESS_[RWX]_OK can go. [ 25986] ftrwrite, ftrexec, fteread, ftewrite and fteexec can all be merged with Perl_pp_ftrread(). [ 25987] Oops. This new panic was meant to be added to the documented diagnostics. [ 25988] Meta-oops. Change 25987 was not the change you were looking for. [ 25989] Once more, with feeling. [Mmm, has someone been swapping the coffee for decaf?] [Actual cause - make sure the edited version of the file is in the correct directory] [ 25990] Merge msgsnd, msgrcv with Perl_pp_shmwrite(). "Perfection is achieved not when there's nothing more to add, but when there's nothing left to remove" (Antoine de Saint-Exupéry) [ 25991] Subject: Re: Configure -Dno_mathoms From: gisle@ActiveState.com (Gisle Aas) Date: 01 Nov 2005 03:27:49 -0800 Message-ID: [ 25999] The remaining special logic in pp_syswrite can be moved into pp_send, which is actually already 50% syswrite. [ 26000] pp_semop can fit within Perl_pp_shmwrite(). [ 26053] Subject: patch@26052 - compile rule for mathoms.c From: "John E. Malmberg" Date: Tue, 08 Nov 2005 23:09:03 -0500 Message-ID: <4371765F.1000900@qsl.net> Branch: maint-5.8/perl !> mathoms.c opcode.h opcode.pl perl.c pod/perldiag.pod pp_sys.c !> vms/descrip_mms.template ____________________________________________________________________________ [ 27314] By: nicholas on 2006/02/24 14:21:24 Log: Integrate: [ 25929] It appears that it's trivially easy to alias opcode functions in the opcode table. We should be able to get both a speed and size saving here. Time will tell how much. [ 25930] Try to preserve the lost opcode alias info in comments extent to change #25929 [ 25937] Replace the 3 currently "unreachable" ops with a single op body unimplemented_op() that panics descriptively if ever reached. [ 25939] Simplify the implementation of the "unimplemented" message in the conditionally compiled ops. [ 25940] All the ops that are simply C can be bypassed in the op dispatch table, and the bodies retired. [ 25958] Merge ftsize ftmtime ftatime ftctime into Perl_pp_ftis [ 25959] Correct error in 25958 for threaded builds. [ 25960] Avoid a pad panic by attempting to use dTARGET; in an op that didn't flag that it needed a target (OP_FTIS) [ 25961] Merge ftzero ftsock ftchr ftblk ftfile ftdir ftpipe into Perl_pp_ftrowned [ 25962] Move ftsuid ftsgid ftsvtx into Perl_pp_ftrowned. [ 25963] Move the conditonal compiles that report absence of kill and chown from pp_kill() and pp_chown() into apply(). [ 25964] unlink chmod utime kill can be merged into Perl_pp_chown(). [ 25968] Avoiding source code duplication for the lstat() on filehandle %s warning also makes the object code slightly smaller. [ 25972] Perl_pp_symlink and Perl_pp_link can be merged. The diff looks evil, but the actual finished code is not as bad as it seems. [ 25982] Subject: [PATCH] bad semicolon in pp_ftrowned From: Andy Lester Date: Fri, 4 Nov 2005 00:37:54 -0600 Message-ID: <20051104063753.GA3863@petdance.com> [ 25997] Fix a Win32 linker error following change #25972 Branch: maint-5.8/perl !> doio.c mathoms.c opcode.h opcode.pl pod/perldiag.pod pp.c !> pp_ctl.c pp_sys.c ____________________________________________________________________________ [ 27310] By: nicholas on 2006/02/24 13:20:45 Log: Integrate: [ 25893] Replace gv_fetchmethod() with a macro to call gv_fetchmethod_autoload() with the extra TRUE argument. [ 25895] Replace hv_iternext() with a macro that calls hv_iternext_flags with an extra 0 argument. Move the old body to mathoms.c [ 25897] Replace hv_magic() with a macro to call sv_magic() directly. Move the old body to mathoms.c [ 25898] init_i18nl14n is a mathom. [ 25900] save_hints is a mathom. [ 25901] Functions that die aren't exactly well used code. (ck_retarget, oopsCV, pp_padany, pp_threadsv, pp_mapstart) [ 25903] Replace is_utf8_string_loc() with a macro that passes the extra 0 argument to is_utf8_string_loc(). Correct the description of its parameters in its POD. [ 25905] Replace uvuni_to_utf8() with a macro that passes the extra 0 argument to uvuni_to_utf8_flags(). Move the old body to mathoms.c [ 25906] Given that sv_nosharing performs the same function as sv_nolocking and sv_unnolocking (ie sweet FA), we might as well use the 1 function to initialise all 3 variables, and elimiate the other two. For some reason all 3 are listed as being in the public API. Daft. [ 25907] Perl_is_utf8_* share a lot of common code. Pull that out into a new function S_is_utf8_common. [ 25909] is_utf8_alnum() and is_utf8_alnumc() can use is_utf8_common() too. [ 25910] const const bad bad. gcc bad bad too, because it didn't grumble one bit. (or two, for that matter). [ 25911] perlsio_binmode() is pretty much a mathom on UNIX platforms, but it is used on Cygwin, at least. [ 25916] is_utf8_string_loc() is now a macro, don't use its Perl_-prefixed form [ 25918] Fixed threaded builds following change 25916 [ 25921] uvchr_to_utf8() and utf8n_to_uvchr() are mathoms on ASCII based systems, and not on EBCDIC, so some more thinking is going to be needed here. [ 25926] A more elegant way to deal with utf8n_to_uvchr() and utf8n_to_uvuni(). [ 25946] This should clear up 'Perl_do_exec' undefined; warnings on win32 [ 25947] This *really* should clear up Win32's Perl_do_exec undefined warnings Branch: maint-5.8/perl ! global.sym mathoms.c proto.h utf8.c !> embed.fnc embed.h gv.c gv.h hv.c hv.h intrpvar.h locale.c op.c !> perl.h perlio.c pp.c pp_ctl.c pp_hot.c utf8.h util.c ____________________________________________________________________________ [ 27308] By: nicholas on 2006/02/24 12:11:35 Log: Integrate: [ 25883] sv_2pv_nolen, sv_2pvbyte_nolen and sv_2pvutf8_nolen can all be replaced with terse macros, and the function bodies retired. [ 25884] Add a new SMAGIC flag, to signal a call to SvSETMAGIC. Add it to sv_catpvn_flags and sv_catsv_flags, and then re-implement sv_catpvn_mg and sv_catsv_mg as calls to sv_catpvn_flags and sv_catsv_flags respectively. [ 25885] Replace sv_catsv_mg and sv_catpvn_mg, and move the obsolete bodies to mathoms.c [ 25886] Replace the body of the rarely used sv_setpviv_mg with a call to sv_setpviv followed by SvSETMAGIC. [ 25889] Replace do_open() with a macro to call do_openn() with the extra two 0 paramters. Move the old body to mathoms.c. [ 25890] Replace do_exec() with a macro to call do_exec3() with the extra two 0 arguments. Move the old body to mathoms.c [ 25891] Replace do_aexec() with a macro to call do_aexec5() with the two extra 0 arguments. Move the old body to mathoms.c [ 25892] I think we have a winner. do_pipe hasn't been called since 4.036. Maybe we should just remove this appendix. [ 25896] Fix win32 breakage inadvertently introduced by 25889 [ 25899] I infer that change 25891 as-is would break OS2. This should fix it. Branch: maint-5.8/perl ! doio.c mathoms.c !> embed.fnc embed.h perl.h proto.h sv.c sv.h ____________________________________________________________________________ [ 27305] By: nicholas on 2006/02/24 11:35:40 Log: Integrate: [ 25867] Makefiles like tabs, not spaces. [ 25869] gv_fullname and gv_efullname are definately mathoms. [ 25871] gv_fullname3 and gv_efullname3 are definately mathoms. They even say so. [ 25872] Coverage stats say av_fake() is a mathom too. [ 25873] Additional mathoms from doio.c [ 25875] Change to use sv_force_normal_flags() directly, rather than calling sv_force_normal() to call sv_force_normal_flags(). [ 25876] sv_taint() can easily be replaced by a macro. [ 25877] Oops, missed an sv_force_normal() [ 25878] Replace sv_unref with a macro that calls sv_unref_flags [ 25879] Replace sv_force_normal with a macro that calls sv_force_normal_flags. [ 25880] sv_utf8_upgrade declares itself to be a mathom, so off it goes. Remove duplicate declaration of Perl_sv_2uv [ 25882] All the private implementations of @foo for compilers that can't do complex macro expansions became dead wood sometime around 5.7.3 (Don't tell Otava this) Branch: maint-5.8/perl ! mathoms.c !> NetWare/sv_nw.c av.c doio.c embed.fnc embed.h gv.c mg.c !> proto.h sv.c sv.h win32/Makefile win32/makefile.mk ____________________________________________________________________________ [ 27303] By: nicholas on 2006/02/24 10:56:28 Log: Integrate: [ 25866] Add a new file, mathoms.c, to hold old code kept around for binary compatability with previous versions of Perl. Change also includes various Makefile changes to compile the new file and link it into libperl. (Would have pulled more, but perforce branching behaviour would bite) Branch: maint-5.8/perl + mathoms.c !> MANIFEST Makefile.SH Makefile.micro NetWare/Makefile globals.c !> numeric.c op.c perlio.c plan9/mkfile sv.c utf8.c !> vms/descrip_mms.template win32/Makefile win32/makefile.mk !> wince/Makefile.ce ____________________________________________________________________________ [ 27285] By: nicholas on 2006/02/23 15:02:49 Log: Update Changes Branch: maint-5.8/perl ! Changes patchlevel.h ____________________________________________________________________________ [ 27284] By: nicholas on 2006/02/23 14:49:22 Log: Integrate: [ 27002] Code assumes that *FOO{SCALAR} will always return a scalar reference, so make it so, creating $FOO if necessary. (Effectively this is a policy decision that PERL_DONT_CREATE_GVSV is visible to XS code, but isn't visible to Perl code) Branch: maint-5.8/perl !> pp.c t/op/gv.t ____________________________________________________________________________ [ 27281] By: nicholas on 2006/02/23 13:35:27 Log: Update Changes Branch: maint-5.8/perl ! Changes patchlevel.h ____________________________________________________________________________ [ 27280] By: nicholas on 2006/02/23 13:32:51 Log: Missed file from running embed.pl Branch: maint-5.8/perl ! embedvar.h ____________________________________________________________________________ [ 27277] By: nicholas on 2006/02/23 00:51:58 Log: Integrate: [ 26980] PERL_DONT_CREATE_GVSV wasn't as effective as it should have been. plus restore the description (but now to perl589delta.pod) Branch: maint-5.8/perl ! pod/perl589delta.pod !> gv.c ____________________________________________________________________________ [ 27275] By: nicholas on 2006/02/22 23:58:35 Log: In Perl_sv_dup, merge the copy code for the 3 simple types (IV,NV,RV). Branch: maint-5.8/perl ! sv.c ____________________________________________________________________________ [ 27266] By: nicholas on 2006/02/21 23:57:15 Log: Integrate: [ 27030] Subject: [patch] arena rework - unify arenaroots From: Jim Cromie Message-ID: <43DF3D0A.3040102@gmail.com> Date: Tue, 31 Jan 2006 03:33:46 -0700 [ 27262] PL_body_arenas should be set to NULL when the interpreter is cleared. Branch: maint-5.8/perl ! sv.c !> embedvar.h hv.c intrpvar.h perlapi.h ____________________________________________________________________________ [ 27253] By: nicholas on 2006/02/20 23:42:25 Log: Integrate: [ 26170] Remove all the now unused new_XFOO()/del_XFOO() macros [ 26172] With -DPURFIY we change the flags so that everything is allocated directly. (Don't use arenas at all.) This seems a rather elegant way to simplify some of the code in sv.c by removing conditional compilation. [ 26173] Norman observed that a couple of macros could make the body defintion table a lot more readable. [ 26184] Flip the sign of the value in body details offset, and change its type to size_t. [ 26439] Avoid negating an unsigned value. (The offset in the SV body table) [ 27090] Eliminate a croak we can't get to. [ 27091] Merge and refactor the PVAV and PVHV code in Perl_sv_upgrade, which reduces the object size by about 80 bytes. [ 27092] Remove "Can't upgrade to undef" from perldiag, as the corresponding code is now gone from sv_upgrade. [ 27094] new_body_type doesn't need to subtract the offset, that's what new_body_allocated is for. [ 27095] The message "Can't upgrade that kind of scalar" can be far more informative. [ 27109] In Perl_sv_upgrade, the assignment to new_type_details only needs to be done once. Branch: maint-5.8/perl ! sv.c !> pod/perldiag.pod ____________________________________________________________________________ [ 27246] By: nicholas on 2006/02/20 16:41:57 Log: Integrate: [ 26169] sv_clear can manipulate the arena array directly too. Also, I think that we may have been "leaking" NV bodies in blead since 25051. Oops. (Will not have been detectable except with -DPURIFY as the leak is not returning them as free to the arena) Branch: maint-5.8/perl ! sv.c ____________________________________________________________________________ [ 27232] By: nicholas on 2006/02/19 23:57:27 Log: Integrate: [ 26165] Eliminate new_body_offset from sv_dup() [ 26166] Eliminate new_body_length from sv_dup. Replace SvTYPE(sstr) with sv_type, which is a constant variable with the same value. [ 26167] Merge the arms of the switch. The decision about what comes from an arena and what comes from malloc is now solely in the table. [ 26168] We only need to zero new bodies for upgrades. sv_dup doesn't need to zero as it's copying over the same sized structure. So change new_NOARENA to malloc(), and add new_NOARENAZ() to calloc() [ 26378] Perl_sv_dup should be allocating bodies based on *size*, not the length to copy. [ 27159] Coalesce 2 if blocks with duplicated conditions in Perl_sv_dup. Branch: maint-5.8/perl ! sv.c ____________________________________________________________________________ [ 27226] By: nicholas on 2006/02/19 12:07:56 Log: Integrate: [ 26164] Merge the last remaining case duplication in sv_upgrade Branch: maint-5.8/perl !> sv.c ____________________________________________________________________________ [ 27225] By: nicholas on 2006/02/19 11:49:31 Log: Integrate: [ 26159] Add a flag to the body details of which types come from arenas. Update the allocation information for all the types. Add a new macro new_NOARENA() for allocating the non-arena types. Use calloc rather than malloc for the new_NOARENA so that the Zero() can be skipped. [ 26161] new_body_inline doesn't need to be passed the root, as it's already inferring the arenaroot directly from the passed-in type. [ 26162] Eliminate new_body_offset from sv_upgrade [ 26163] Eliminate new_body_length from sv_upgrade Branch: maint-5.8/perl ! sv.c ____________________________________________________________________________ [ 27224] By: nicholas on 2006/02/19 11:30:33 Log: Integrate: [ 26155] Rename sv_upgrade's parameter mt to new_type. [ 26156] const and static for the const static private table. Replace the "upgrade" special case for SVt_PV with assertions as it's not been needed since we forced *up*grades. [ 26158] Simplify the expression for the upgrade logic for SVt_IV. Replace the "can't happen" comment for SVt_PVIV with assertions that it didn't happen. (except the bit about assertions) Branch: maint-5.8/perl !> sv.c ____________________________________________________________________________ [ 27217] By: nicholas on 2006/02/18 11:50:06 Log: Integrate: [ 26151] "Can you see what it is yet?" Next steps towards making as much as possible table driven. [ 26152] All that can be in the first switch statement of sv_upgrade is now table driven. The shrinking continues on the morrow. [ 26623] If NV_ZERO_IS_ALLBITS_ZERO is not true, then it's important to inialise the NV only if the new body has an NV. (D'oh!) Branch: maint-5.8/perl !> sv.c ____________________________________________________________________________ [ 27216] By: nicholas on 2006/02/18 11:07:18 Log: Integrate: [ 26146] s/SvTYPE(sv)/old_type/ in sv_upgrade because we already have that value in that variable. Don't compile S_new_body with -DPURIFY [ 26149] Merge sizeof_body_by_svtype and offset_by_svtype into a single array of structures. [ 26150] Add the size of the structure to copy as a field in the per type data table. Branch: maint-5.8/perl ! sv.c ____________________________________________________________________________ [ 27206] By: nicholas on 2006/02/17 13:04:04 Log: Integrate: [ 26141] Subject: Re: eliminate discreet arenaroots From: Jim Cromie Message-ID: <4378E5B0.3010708@gmail.com> Date: Mon, 14 Nov 2005 12:29:52 -0700 [ 26142] PL_pte_root and PL_pte_arenaroot can be exterminated. Which reveals an erroneous remaining reference to PL_pte_root to correct. [ 26143] Oi gcc, No! sizeof(void) is an error. *Not* 1. Wanting sizeof(void) isn't even a compilation error with -ansi -pedantic. Just a warning. Lame! I'm going to send it back for a full refund. Hateful software. [ 26148] Move the location of the definition of the arena slot used for pte into sv.h, to keep it in one central place. Change it to use SVt_RV. (Having SVt_IV for real use will be useful. SVt_RV has no body, and doesn't pretend to have one, so it is spare.) [ 26171] Map the HE arena onto SV type 0 (SVt_NULL). Abolish PL_he_root and PL_he_arenaroot. Branch: maint-5.8/perl ! hv.c intrpvar.h sv.c sv.h !> embedvar.h ext/XS/APItest/APItest.xs perlapi.h ____________________________________________________________________________ [ 27191] By: nicholas on 2006/02/15 09:44:22 Log: Integrate: [ 26129] Provide a crude way for specifying the command line for make test.valgrind [ 26846] Adapt test.valgrind make target to recent valgrinds (3.x) by Jim Cromie [ 27128] Subject: [patch] several broken make *.valgrind targets From: Jim Cromie Date: Tue, 07 Feb 2006 15:11:54 -0700 Message-ID: <43E91B2A.4060700@gmail.com> Branch: maint-5.8/perl ! t/TEST !> Makefile.SH ____________________________________________________________________________ [ 27183] By: nicholas on 2006/02/14 22:06:00 Log: Unlike 5.9, SVt_IV needs to be freed. Only SVt_NULL can exit early. Branch: maint-5.8/perl ! sv.c ____________________________________________________________________________ [ 27175] By: nicholas on 2006/02/13 21:02:41 Log: Getting the old_length wrong for SVt_PVNV in Perl_sv_upgrade causes surprisingly subtle bugs. Now fixed. (Cause being that the expression to get the length of PVNV differs between blead and maint, because blead's structures have been space optimised.) Branch: maint-5.8/perl ! sv.c ____________________________________________________________________________ [ 27170] By: nicholas on 2006/02/13 15:32:10 Log: Integrate: [ 27133] lastparen in the regexp structure is never initialised, and so will still be uninitialised after a failed match, yet various points in mg.c read it come what may. Should fix bug 38461. [ 27169] All the bits of clone_params.flags need to be initialised, not just CLONEf_JOIN_IN. Branch: maint-5.8/perl !> ext/threads/threads.xs regcomp.c ____________________________________________________________________________ [ 27168] By: nicholas on 2006/02/12 21:10:26 Log: Integrate: [ 25904] global.sym appears to have become out of date. Oops. [ 25923] Arrange to only output each symbol once into global.sym [ 25924] Perl_csighandler was in global.sym twice. Branch: maint-5.8/perl !> embed.pl global.sym ____________________________________________________________________________ [ 27167] By: nicholas on 2006/02/12 20:48:31 Log: Make cross upgrades work (eg a request to "upgrade" PV to NV) Branch: maint-5.8/perl ! sv.c ____________________________________________________________________________ [ 27165] By: nicholas on 2006/02/12 18:18:04 Log: Integrate: [ 25051] Unwrap the del_Xfoo() macros by explicitly noting the arena to free the body to. This comletely eliminates the second switch statement. Branch: maint-5.8/perl ! sv.c ____________________________________________________________________________ [ 27164] By: nicholas on 2006/02/12 11:20:06 Log: Perl_sv_upgrade is not void in maint, so need to return TRUE. Branch: maint-5.8/perl ! sv.c ____________________________________________________________________________ [ 27163] By: nicholas on 2006/02/12 00:29:55 Log: Integrate: [ 24897] Move the offset calculations outside of new_body/del_body. This makes some of the upgrade code slightly simpler. [ 24899] If NV_ZERO_IS_ALLBITS_ZERO there's no need to explicitly set the NVX to 0.0, as the memset has already done it. [ 24900] S_del_body is sufficiently small that inlining it is a space win. [ 24918] As there will be no old body to dispose of, we can return immediately. [ 25053] Copying the first parameter in the macro del_body seems to reduce code size. Presumably an expression is no longer evaluated twice. [ 25066] Inlining del_HE is actually a space optimisation. It's therefore likely also to be a speed optimisation. :-) [ 25069] Provide a macro version of S_new_body to inline it within the hot code. [ 25113] Subject: [patch: sv.c] reuse new_body_inline in Perl_new_body From: Jim Cromie Date: Sun, 10 Jul 2005 15:59:30 -0600 Message-ID: <42D19A42.8040409@divsol.com> Branch: maint-5.8/perl !> embed.fnc embed.h hv.c proto.h sv.c ____________________________________________________________________________ [ 27160] By: nicholas on 2006/02/11 15:13:04 Log: Integrate: [ 24896] Refactor sv_dup to memcpy from source to destination, and only explicitly copy the structure members that need explicit duplication. Branch: maint-5.8/perl ! sv.c ____________________________________________________________________________ [ 27157] By: nicholas on 2006/02/11 10:48:48 Log: Integrate: [ 24894] Subject: [PATCH] squeeze repetition out of Perl_sv_free_arenas From: Jim Cromie Message-ID: <42B3C388.2090902@divsol.com> Date: Sat, 18 Jun 2005 00:47:36 -0600 [ 24895] Squeeze a little more savings by moving pte and he arenas into the common arena free code. [ 24914] Replace the non-const initialiser block + loop with a series of calls to a small static function. c.f. "I will replace you with a small shell script..." Branch: maint-5.8/perl !> hv.h perl.h sv.c ____________________________________________________________________________ [ 27156] By: nicholas on 2006/02/11 10:09:35 Log: Integrate: [ 24892] sv_upgrade by memcpy AV and HV cases need tidyup [ 24893] update AVs and HVs using the old_body, and remove all of the local variables used to hold the SV body piecemeal. Branch: maint-5.8/perl ! sv.c ____________________________________________________________________________ [ 27150] By: nicholas on 2006/02/10 11:49:32 Log: Integrate: [ 24886] Move freeing the old body after the creating of the new body. [ 24887] Missing aTHX_ (noticed by Jim Cromie) [ 24888] Collect a little more information about the body we're getting rid of [ 24890] Change 24886 was buggy - should be taking (and passing in) the address of the arena. [ 24891] Fix compiling with -DPURIFY Branch: maint-5.8/perl !> sv.c ____________________________________________________________________________ [ 27149] By: nicholas on 2006/02/10 11:25:06 Log: Integrate: [ 24881] new_pte/del_pte can use new_body/del_body too [ 24885] The only way is up! (*up*grades. Or croak) Branch: maint-5.8/perl ! sv.c ____________________________________________________________________________ [ 27147] By: nicholas on 2006/02/10 11:04:46 Log: Integrate: [ 24878] Replace all the new/del static functions with calls to a single pair of functions that uproot/plant bodies Branch: maint-5.8/perl ! sv.c !> embed.fnc embed.h proto.h ____________________________________________________________________________ [ 27146] By: nicholas on 2006/02/10 10:29:10 Log: Integrate: [ 24876] Replace all the S_more_* functions with a single function. Saves 1.5K on the object file in OS X when built with -Os [ 24877] Making more_* return a pointer, and swapping to a ternary ?: in new_* seems to generate slightly terser object code. Branch: maint-5.8/perl ! sv.c ____________________________________________________________________________ [ 27144] By: nicholas on 2006/02/09 22:33:56 Log: Integrate: [ 24875] Subject: [PATCH] Re: [perl #36048] Refactor S_more_*v into one function From: Jim Cromie Message-ID: <42ADC8ED.4090009@divsol.com> Date: Mon, 13 Jun 2005 11:57:01 -0600 Plus fixups - xiv_allocated, xnv_allocated, xpviv_allocated structures, cast corrections in S_{new,del}_x[in]v, and more_thingy declarations for the 2 types that no longer use arenas in 5.9 (IVs and RVs) Branch: maint-5.8/perl ! embedvar.h intrpvar.h sv.c sv.h ____________________________________________________________________________ [ 27143] By: nicholas on 2006/02/09 21:49:46 Log: Restore changes 24459 and 24467 (GVs from arenas) (ie reverse 26559) as the arena re-working is started. Branch: maint-5.8/perl ! embed.fnc embed.h embedvar.h intrpvar.h perlapi.h proto.h sv.c ____________________________________________________________________________ [ 27142] By: nicholas on 2006/02/09 21:12:42 Log: Integrate: [ 24544] create an "allocated" structure for PVs, PVAVs and PVHVs Branch: maint-5.8/perl !> av.h hv.h intrpvar.h sv.c sv.h ____________________________________________________________________________ [ 27141] By: nicholas on 2006/02/09 20:34:26 Log: Integrate: [ 24258] Merge common code. Reduces object size by 1K on x86, 6K on PPC [ 24340] The if() clause is the wrong way round, so OOK can never be true in this case. Given that OOK is never undone, we could never have been correctly upgrading a PV with OOK to HV or AV. I believe that only SVt_NULL is the only type ever upgraded to HV or AV. Branch: maint-5.8/perl ! sv.c ____________________________________________________________________________ [ 27058] By: nicholas on 2006/02/03 11:35:30 Log: Create perl589delta.pod Branch: maint-5.8/perl + pod/perl589delta.pod ! MANIFEST Makefile.SH pod.lst pod/perl.pod ! vms/descrip_mms.template win32/Makefile win32/makefile.mk ! win32/pod.mak ____________________________________________________________________________ [ 27055] By: nicholas on 2006/02/02 17:28:37 Log: Update Changes, and create Changes5.8.8 Branch: maint-5.8/perl +> Changes5.8.8 ! Changes MANIFEST patchlevel.h ____________________________________________________________________________ [ 27041] By: nicholas on 2006/02/01 18:22:12 Log: That was 5.8.8. Branch: maint-5.8/perl ! patchlevel.h ____________________________________________________________________________ [ 27040] By: nicholas on 2006/02/01 18:08:07 Log: Regenerated files that shipped with 5.8.8. Branch: maint-5.8/perl ! ext/B/B/Asmdata.pm perlapi.c ____________________________________________________________________________ [ 27029] By: nicholas on 2006/01/31 23:39:33 Log: lib/CPAN/SIGNATURE is supposed to be in the MANIFEST, and therefore the tarball. Branch: maint-5.8/perl ! MANIFEST ____________________________________________________________________________ [ 27026] By: nicholas on 2006/01/31 16:25:24 Log: Break a leg. Branch: maint-5.8/perl ! patchlevel.h ____________________________________________________________________________ [ 27025] By: nicholas on 2006/01/31 16:03:49 Log: Update Changes Branch: maint-5.8/perl ! Changes patchlevel.h ____________________________________________________________________________