This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
David Mitchell [Tue, 10 Nov 2015 13:50:51 +0000 (13:50 +0000)]
[MERGE] faster arithmetic
David Mitchell [Sun, 25 Oct 2015 18:28:14 +0000 (18:28 +0000)]
split pp_postdec() from pp_postinc() and improve
pp_postinc() handles both $x++ and $x-- (and the integer variants
pp_i_postinc/dec). Split it into two separate functions, as handling
both inc and dec in the same function requires 3 extra conditionals.
At the same time make the code more efficient.
As currently written it:
1) checked for "bad" SVs (such as read-only) and croaked;
2) did a sv_setsv(TARG, TOPs) to return a copy of the original value;
2) checked for a IOK-only SV and if so, directly incremented the IVX slot;
3) else called out to sv_inc/dec() to handle the more complex cases.
This commit combines the checks in (1) and (3) into one single big
check of flags, and for the simple integer case, skips 2) and does
a more efficient SETi() instead.
For the non-simple case, both pp_postinc() and pp_postdec() now call a
common static function to handle everything else.
Porting/bench.pl shows the following raw numbers for
'$y = $x++' ($x and $y lexical and holding integers):
before after
------ -----
Ir 306.0 223.0
Dr 106.0 82.0
Dw 51.0 44.0
COND 48.0 33.0
IND 8.0 6.0
COND_m 1.9 0.0
IND_m 4.0 4.0
David Mitchell [Sun, 25 Oct 2015 08:41:50 +0000 (08:41 +0000)]
split pp_predec() from pp_preinc() and improve
pp_preinc() handles both ++$x and --$x (and the integer variants
pp_i_preinc/dec). Split it into two separate functions, as handling
both inc and dec in the same function requires 3 extra conditionals.
At the same time make the code more efficient.
As currently written it:
1) checked for "bad" SVs (such as read-only) and croaked;
2) checked for a IOK-only SV and directly incremented the IVX slot;
3) else called out to sv_inc() to handle the more complex cases.
This commit combines the checks in (1) and (2) into one single big
check of flags, and anything "bad" simply skips the IOK-only code
and calls sv_dec(), which can do its own checking of read-only etc
and croak if necessary. Porting/bench.pl shows the following raw numbers
for ++$x ($x lexical and holding an integer):
before after
-------- --------
Ir 77.0 56.0
Dr 30.0 24.0
Dw 10.0 10.0
COND 12.0 9.0
IND 2.0 2.0
COND_m -0.1 0.0
IND_m 2.0 2.0
Even having split the function into two, the combined size of the two new
functions is smaller than the single previous function.
David Mitchell [Thu, 22 Oct 2015 11:04:40 +0000 (12:04 +0100)]
faster add, subtract, multiply
In pp_add, pp_subtract and pp_multiply, special-case the following:
* both args IV, neither arg large enough to under/overflow
* both args NV.
Starting in 5.8.0, the implementation of the arithmetic pp functions
became a lot more complex (and famously, much slower), due to the need
to support 64-bit integers.
For example, formerly pp_add just converted both its args to an NV and
returned an NV. On 64-bit systems, that could gave bad results if the
mantissa of an NV was < 64 bits; for example:
$ perl561 -e'$x = 0x1000000000000000; printf "%x\n", $x+1'
1000000000000000
$ perl580 -e'$x = 0x1000000000000000; printf "%x\n", $x+1'
1000000000000001
This led to a lot of complex code that covered all the possibilities
of overflow etc.
This commit adds some special casing to these three common arithmetic ops.
It does some quick checks (mainly involving fast boolean and bit ops)
to determine if both args are valid IVs (and not UVs), are not magic,
and aren't very big (+ve or -ve). In this case, the result is simply
SvIVX(svl) + SvIVX(svr) (or - or *) with no possibility of overflow.
Failing that, if both args are NV's and not magic, then if both NVs
can be converted to IVs without loss, handle as for the IV case; failing
that, just return SvNVX(svl) + SvNVX(svr);
For all other cases, such as mixed IV and NV or PV, fall back to the old
code.
On my platform (x86_64), it (along with the previous commit) reduces the
execution time of the nbody benchmark (lots of floating-point vector
arithmetic) by a third and in fact makes it 10% faster than 5.6.1.
David Mitchell [Thu, 22 Oct 2015 15:43:49 +0000 (16:43 +0100)]
make SETi/u/n, (X)PUSHi/u/n more efficient
These macros are used by many pp functions to set TARG to an int or float
value and put it on the stack. The macros use a call to sv_setiv()
or similar.
However, there is a good chance that the target is ether a lexical var
or a PADTMP that has been assigned just an IV/NV in the past, in which
case its body is likely to still be of type SVt_IV/SVt_NV. If this is
the case, we can set its value much more efficiently.
Update those setting macros to first check if the target has a simple
body, and if so, set directly.
This makes quite a significant performance difference on code that does
a lot of arithmetic, at the cost of a larger binary (0.36% on my Linux
x86_64 system).
It also allows you (at compile time) to skip testing for taint if you
know that the value can't be tainted (e.g. pp_add() when both args are
non-magical and so can't have taint magic attached). The next commit will
make use of this.
Includes a couple of efficiency tweaks suggested by Daniel Dragan.
David Mitchell [Fri, 30 Oct 2015 13:44:11 +0000 (13:44 +0000)]
avoid (TAINTING_get && TAINT_get)
In various places we test for both (PL_tainting && PL_tainted).
Since if tainting isn't enabled PL_tainted should never get set, it's
more efficient to just test for (TAINT_get).
We ensure that PL_tainted doesn't actually get set when !PL_tainting
by changing some "setting" macros from PL_tainted = TRUE to
PL_tainted = PL_tainting.
Ricardo Signes [Mon, 9 Nov 2015 22:40:51 +0000 (17:40 -0500)]
base: prepare to make a new CPAN release
...eliminates some unneeded files, adds a Makefile.PL (for the
INSTALLDIRS-picking behavior).
Reini Urban [Mon, 16 Mar 2015 09:27:37 +0000 (10:27 +0100)]
MARK -Ds debugging
display the MARK arity and pointers with MARK macros.
assert on markptr underflow.
Aaron Crane [Tue, 10 Nov 2015 01:22:13 +0000 (01:22 +0000)]
perlexperiment.pod: use consistent style for Perl versions
Aaron Crane [Tue, 10 Nov 2015 01:16:28 +0000 (01:16 +0000)]
Mention warning categories for removed experimental features
This seems likely to be helpful for future code archæologists.
The string "autoderef" string didn't previously appear in perlexperiment.pod;
that looks like an oversight.
Dagfinn Ilmari Mannsåker [Mon, 9 Nov 2015 16:51:47 +0000 (16:51 +0000)]
Update perlexperiment for removal of lexical topic and autoderef
Karl Williamson [Mon, 9 Nov 2015 05:59:51 +0000 (22:59 -0700)]
t/op/utf8decode.t: print debugging info if test fails
Karl Williamson [Thu, 29 Oct 2015 15:27:48 +0000 (09:27 -0600)]
utf8.h: Move #define within file
This #define has, until the next commit, not been needed on EBCDIC
platforms; move it in preparation for that commit.
Karl Williamson [Fri, 30 Oct 2015 02:32:08 +0000 (20:32 -0600)]
utf8.h, utfebcdic.h: Use mnemonic constant
The magic number 13 is used in various places on ASCII platforms, and
7 correspondingly on EBCDIC. This moves the #defines for what these
represent to early in their files, and uses the symbolic name
thereafter.
Karl Williamson [Fri, 30 Oct 2015 02:24:14 +0000 (20:24 -0600)]
utf8.h: Reformat a couple of table lines
This is just an aesthetic issue, hopefully making things clearer.
Karl Williamson [Sun, 25 Oct 2015 02:08:47 +0000 (20:08 -0600)]
re/pat_advanced.t: Fix a test so also runs on EBCDIC
Karl Williamson [Sun, 25 Oct 2015 02:04:17 +0000 (20:04 -0600)]
Devel::Peek: document that uses STDERR
Tony Cook [Thu, 29 Oct 2015 04:07:09 +0000 (15:07 +1100)]
[perl #126469] document the ob parameter to sv_reftype()
Tony Cook [Thu, 29 Oct 2015 04:05:59 +0000 (15:05 +1100)]
make sv_ref() part of the API
The existing sv_reftype() returns the class of a blessed SV only
as a const char *, so we can't tell if it's UTF8 or not.
This probably should have been API from the beginning.
Tony Cook [Mon, 12 Oct 2015 03:02:33 +0000 (14:02 +1100)]
cygwin: look harder for a C: mapping and skip if we don't find it
Some recent update to df on cygwin changed the output from listing
all mappings for the drive with cygwin installed on it to listing
only the root directory, eg from:
$ df
Filesystem 1K-blocks Used Available Use% Mounted on
C:/cygwin/bin
76943280 45410396 31532884 60% /usr/bin
C:/cygwin/lib
76943280 45410396 31532884 60% /usr/lib
C:/cygwin
76943280 45410396 31532884 60% /
C:
76943280 45410396 31532884 60% /cygdrive/c
to:
$ df
Filesystem 1K-blocks Used Available Use% Mounted on
C:/cygwin
1953411068 1261572900 691838168 65% /
which meant the test used an incorrect fallback value.
So force df to display all the mappings, handle the differently
formatted mapping correctly, and if all that fails, skip the test.
Jarkko Hietaniemi [Sat, 7 Nov 2015 19:38:21 +0000 (14:38 -0500)]
perl #126586 hexfp may lose 1-3 low order bits (most often, 1)
Jarkko Hietaniemi [Fri, 6 Nov 2015 23:54:16 +0000 (18:54 -0500)]
perl #126582 hexfp overflow drops hi-order bits
Ricardo Signes [Fri, 6 Nov 2015 14:54:52 +0000 (09:54 -0500)]
Carp: stable CPAN release
Karl Williamson [Sun, 25 Oct 2015 02:03:31 +0000 (20:03 -0600)]
Dumpvalue: Generalize for non-ASCII platforms
I overlooked this module until now. It turns out that much of the code
I had changed had a common ancestor with the code I had already changed
to work on non-ASCII platforms in lib/dumpvar.pl. So I just copied
that, changing the things that needed to be different.
It appears that Dumpvalue had a bug, in that it did not escape NUL, of
all the C0 controls. I changed it to do so.
Steve Hay [Thu, 5 Nov 2015 17:36:11 +0000 (17:36 +0000)]
Update other win32/ makefiles as per
eb840d4ab6
Daniel Dragan [Wed, 4 Nov 2015 22:28:06 +0000 (17:28 -0500)]
remove useless build product /win32/config.w32
CFGSH_TMPL (config.gc or config.vc) is not written to by config_sh.PL, so
there is no need to make a copy of it, this reduces 1 target node out of
dmake so there are less targets to traverse for dep checking, this also
fixes an "dmake -n" schedule irregularity where
"copy $(CFGSH_TMPL) config.w32" was scheduled (for serial) as the very
first recipie to run by dmake for a "dmake -n Extensions" but a recipie
in the middle of the procs to launch for "dmake -n Extensions_nonxs". This
schedule irregularity might be causing delays in finding parallel work to
do. Also arrange ..\config.sh target's deps from "on disk" to
"build product" order, so the left side dep nodes get marked "completed"
ASAP in the graph.
--- C:\p523\src\win32\xs.txt
+++ C:\p523\src\win32\nonxs.txt
@@ -1,3 +1,4 @@
+copy config.vc config.w32
if not exist ".\mini" mkdir ".\mini"
if exist config.h del /f config.h
copy config_H.vc config.h
@@ -141,20 +142,11 @@
@C:\DOCUME~1\ADMINI~1\LOCALS~1\Temp\mk31
if exist ..\miniperl.exe.manifest mt -nologo -manifest **CUT**
..\miniperl.exe -I..\lib -f ..\write_buildcustomize.pl ..
-copy config.vc config.w32
..\miniperl.exe -I..\lib config_sh.PL --cfgsh-option-file \
C:\DOCUME~1\ADMINI~1\LOCALS~1\Temp\mk32 config.w32 > ..\config.sh
..\miniperl.exe -I..\lib ..\configpm --chdir=..
xcopy /f /r /i /d /y config.h ..\lib\CORE\*.*
..\miniperl.exe -I..\lib config_h.PL "ARCHPREFIX="
[Committer indented above diff to allow "git am" to apply the patch.]
Daniel Dragan [Thu, 5 Nov 2015 03:34:54 +0000 (22:34 -0500)]
re-parallelize Win32 build after Unicode::Normalize got XS
commit
57dca39807 serialized the 1st and 3rd longest targets (Extensions
and UNIDATAFILES aka mktables) together after commit
c6b7cc2176 which
upgraded Unicode::Normalize, and Unicode::Normalize went from PP->XS,
and the XS version requires unicore/CombiningClass.pl in
cpan/Unicode-Normalize/mkheader during U::N's building. The serialization
added about 32 second to a "dmake -P8 all" on a 8 core machine. Fix this
by spltting off U::N from Extensions and put it in its own target. Making
Extensions, the longest target slightly shorter by removing 1 module
from its workload is another plus. See perl #126564 for time savings details.
[Committer filled in RT ticket number in commit message.]
Chris 'BinGOs' Williams [Tue, 3 Nov 2015 12:56:07 +0000 (12:56 +0000)]
Update Pod-Simple to CPAN version 3.32
[DELTA]
2015-11-02 Marc Green <marcgreen@cpan.org>
* Release 3.32
Fixed failing tests on Windows. Thanks to A. Sinan Unur for the
patch!
Switched debugging output from STDOUT to STDERR. Should rarely be
used, but modules that do depend on debugging output might need to
change how they handle it. Patch from Karl Williamson (GitHub Pull
Request #76).
Added errata_seen() to make POD errors easily accessible. Thanks to
Sean Zellmer for the pull request!
2015-08-23 Marc Green <marcgreen@cpan.org>
* Release 3.31
No changes since 3.30_1.
2015-07-19 Marc Green <marcgreen@cpan.org>
* Release 3.30_1
Simplified the detection of case-insensitivity in Pod::Simple::Search.
Fixed "Use of uninitialized value $1 in lc" warning in
Pod::Simple::Search.
If @INC includes the current directory symbol, '.', the survey()
method of Pod::Simple::Search no longer excludes it from its list
of directories to search. Instead, The survey() and find() methods
now both exclude duplicate directories from @INC (RT #102344).
Moved source repository and updated links to new perl-pod GitHub
organization: https://github.com/perl-pod/pod-simple.
Improved repository links and added GitHub issue tracking link to
the distribution metadata.
Switched from File::Spec's catdir to catfile for path names, to
fix failures on VMS. Also now use Unix path semantics where
they're not required to be platform-specific. Thanks to Craig A.
Berry for the patch (RT #105511).
Improved the example use of the 'html_encode_chars()' method in
the Pod::Simple::XHTML documentation. Patch from Randy Stauner.
Niko Tyni [Sun, 12 Jul 2015 19:23:00 +0000 (22:23 +0300)]
Allow overriding the compile time in "perl -V" output
The C preprocessor macros __DATE__ and __TIME__ embed the compile time
into the binary for the purposes of "perl -V" output. This makes the
build unreproducible: compiling the same source with the same toolchain
cannot be made to yield bitwise identical binaries and other generated
files.
The compile time can now be overridden with the PERL_BUILD_DATE macro.
Bug: https://rt.perl.org/Ticket/Display.html?id=125830
Bug-Debian: https://bugs.debian.org/774422
Patch-Name: debian/do-not-record-build-date.diff
Daniel Dragan [Sat, 31 Oct 2015 10:53:54 +0000 (06:53 -0400)]
remove XSLoader and DynaLoader OS specific code on NA OSes
@dl_resolve_using is only used on dld/freemint, and HPUX shlib loader OSes.
Dont create the array and glob on OSes where @dl_resolve_using is ignored.
sub dl_undef_symbols is only implmented on dld and freemint (freemint is
dld under a different name), otherwise it always returns empty list. Dont
call the sub on OSes where we knows it's constant retval. Saves ops+compile
time for sub bootstrap.
Daniel Dragan [Fri, 30 Oct 2015 05:47:20 +0000 (01:47 -0400)]
remove remains of dl_dld.xs
dl_dld.xs was removed in 5.19.4 in commit
1e5d7f5401
Chris 'BinGOs' Williams [Mon, 2 Nov 2015 13:46:26 +0000 (13:46 +0000)]
Update Math-BigInt-FastCalc to CPAN version 0.35
[DELTA]
2015-10-28 v0.35 pjacklam
* Sync test files with Math-BigInt-1.999707.
* Update the README file.
* Replace 'use vars ...' with 'our ...'. We require a Perl newer than 5.6.0
anyway.
* Required version of Math-BigInt is now 1.999706.
* Move 'Test::More' from 'build_requires' to 'test_requires' in Makefile.PL.
Chris 'BinGOs' Williams [Mon, 2 Nov 2015 13:41:43 +0000 (13:41 +0000)]
Update Getopt-Long to CPAN version 2.48
[DELTA]
Changes in version 2.48
-----------------------
* Fix bug https://rt.cpan.org/Ticket/Display.html?id=39052.
Thanks to Roy Ivy III for digging this out and providing patches.
David Mitchell [Mon, 2 Nov 2015 11:11:17 +0000 (11:11 +0000)]
pp_hot.c: rationalise SvREFCNT_inc_*()
Ensure the correct combinations of _simple, _void, _NN are used.
This fixes a couple of void compiler warnings in pp_aassign.
Steve Hay [Mon, 2 Nov 2015 08:26:02 +0000 (08:26 +0000)]
Upgrade Math::BigInt from version 1.999706 to 1.999707
Daniel Dragan [Sun, 25 Oct 2015 23:31:27 +0000 (19:31 -0400)]
dont install PPPort.so/PPPort.dll
This shared lib is only used for PPPort testing itself, it is similar to
APItest.dll in purpose. PPPort.pm never uses XSLoader/DynaLoader, only its
.t files do. This saves 616KB in the final install dir on Win32, and
atleast one or two dozen KB on all OSes. Since where is auto dir, and what
is arch dir, is complicated and unportable (atleast to me), and what other
files live next to the shared lib (examples, .pdb file, .bs file, .a file)
match the directory fragment, not the files inside of it or the dirs
full path.
Directory of C:\p523\src\lib\auto\Devel\PPPort
10/25/2015 07:16 PM <DIR> .
10/25/2015 07:16 PM <DIR> ..
10/25/2015 07:16 PM 0 .exists
10/25/2015 07:16 PM 59,392 PPPort.dll
10/25/2015 07:16 PM 796 PPPort.exp
10/25/2015 07:16 PM 1,738 PPPort.lib
10/25/2015 07:16 PM 569,344 PPPort.pdb
5 File(s) 631,270 bytes
Jarkko Hietaniemi [Sun, 1 Nov 2015 17:55:49 +0000 (12:55 -0500)]
Workaround for perl #124212: these functions are not true static inline
By "true" I mean that they have prototypes, but no bodies.
So don't declare their prototypes under PERL_NO_INLINE_FUNCTIONS.
After some studying of http://www.greenend.org.uk/rjk/tech/inline.html
it seems like Perl is trying to implement the "simple portable" model.
But the functions listed as failing during porting/extrefs.t in Tru64:
they are neither fish nor fowl. Their prototypes are listed in
proto.h as PERL_STATIC_INLINE (which in Tru64 is "static inline"),
but since the test is built with -DPERL_NO_INLINE_FUNCTIONS,
the function bodies (which would be in inline.h) are not visible.
So they end up being body-less static inline prototypes, which is,
I believe, somewhat of an oxymoron.
The "complicated portable" model might be a more wortwhile longer
term goal: in that, there is no "static inline", and there would be
a new source file, say, inline.c. Now with the "simple portable",
the bodies might end up being compiled multiple times, multiple copies
ending up in different object files, depending on how smart the
compiler/linker is.
Another move could be that maybe there should be no prototypes at all
for inlineables, because having those is kind beside the point. How
well that would work across different compilers is unknown.
Yet another move, perhaps the simplest one, would be to move these
particular functions away from inline.h. But this would be just
dodging the larger problems discussed above.
Lukas Mai [Sun, 1 Nov 2015 12:37:34 +0000 (13:37 +0100)]
restore comma removed by
2f1fe8a307
Lukas Mai [Sat, 31 Oct 2015 19:42:22 +0000 (20:42 +0100)]
fix typo in error message
Steve Hay [Sat, 31 Oct 2015 19:07:14 +0000 (19:07 +0000)]
Add epigraph for 5.22.1-RC1
Steve Hay [Sat, 31 Oct 2015 14:45:43 +0000 (14:45 +0000)]
Perl 5.22.1-RC1 today
Steve Hay [Sat, 31 Oct 2015 14:18:27 +0000 (14:18 +0000)]
Module::CoreList updates for 5.22.1
Craig A. Berry [Sat, 31 Oct 2015 01:58:15 +0000 (20:58 -0500)]
Remove redundant code in configure.com
These lines were made redundant by
054a3baf7ca16fe02 so are just
taking up space.
Jarkko Hietaniemi [Fri, 30 Oct 2015 22:25:51 +0000 (18:25 -0400)]
make regen for uconfig.h
Jarkko Hietaniemi [Fri, 30 Oct 2015 21:50:42 +0000 (17:50 -0400)]
Revert "Remove unused filesystem stat symbols."
This reverts commit
821805a244cacd9869331999cd53407f3323206a.
What's out, is out.
perl #107904 Filesys-Df
perl #108189 Filesys-DfPortable
perl #108191 Filesys-Statvfs
perl #126368 Filesys-DfPortable
Karl Williamson [Wed, 21 Oct 2015 23:04:20 +0000 (17:04 -0600)]
perlre: Nits
This mostly adds C<> formatting, but there are a few updates,
clarifications, and grammar-type fixes.
Karl Williamson [Fri, 30 Oct 2015 04:07:11 +0000 (22:07 -0600)]
PATCH: [perl #126481] panic for !! with syntax error in /(?[...])/
This is fixed by not putting two adjacent '!' operators on the stack.
These are the only right-associative operators in the grammar, and they
just cancel each other out.
Ricardo Signes [Fri, 30 Oct 2015 12:17:15 +0000 (08:17 -0400)]
Carp: CPAN release 1.37_02
Ricardo Signes [Thu, 29 Oct 2015 20:50:03 +0000 (16:50 -0400)]
Carp: remove prereq on parent.pm
Steve Hay [Thu, 29 Oct 2015 13:54:23 +0000 (13:54 +0000)]
Upgrade Math-BigInt from version 1.999705 to 1.999706
Jarkko Hietaniemi [Wed, 28 Oct 2015 11:56:23 +0000 (07:56 -0400)]
For perl #126468: protect quotes in myccflags.
Fixes a problem introduced by
57d2761b where ccflags contents
with quotes got broken by the quotes getting stripped.
Tony Cook [Thu, 29 Oct 2015 03:13:36 +0000 (14:13 +1100)]
Unicode::Normalize needs CombiningClass.pl
This (typically) worked ok for parallel builds, since U::N is built
very late in the XS extensions build process, but for non-parallel
builds the lack of the dependency could result in a build failure.
Jarkko Hietaniemi [Wed, 28 Oct 2015 12:32:33 +0000 (08:32 -0400)]
Unicode::Normalize no-go with miniperl.
Unicode::Normalize comes in via Unicode::UCD, which comes in e.g.
via uni/case.pl.
Tried using test.pl and its skip_all_if_miniperl() in all cases,
but that leads into weird failures in full 'make test'.
Ruud H.G. van Tol [Wed, 28 Oct 2015 14:04:38 +0000 (14:04 +0000)]
Fix Tie::StdScalar when $instance exists but isn't a 'true' value
Patch received from Ruud H.G. van Tol <rvtol@isolution.nl>.
Committer added ++$VERSION and Porting/checkAUTHORS.pl change.
Ricardo Signes [Wed, 28 Oct 2015 15:18:59 +0000 (11:18 -0400)]
document -O change to Configure
Ivan Pozdeev [Wed, 12 Aug 2015 17:33:12 +0000 (20:33 +0300)]
Make -O behaviour the default
David Mitchell [Tue, 20 Oct 2015 15:29:48 +0000 (16:29 +0100)]
pp_sys.c: silence g++ compiler warning
The warning is harmless but annoying
David Mitchell [Tue, 20 Oct 2015 14:04:49 +0000 (15:04 +0100)]
RT: #126309 die more gracefully on (1) x ~1
Recent improvements to MEXTEND() etc means that the above is now caught and
panics rather than crashing. However, the panic is:
panic: av_extend_guts() negative count (-
9223372036854775681)
which is safe, but not pretty. This commit makes it croak instead with:
Out of memory during stack extend
Basically Perl_stack_grow() adds an extra 128 bytes of headroom to the
amount it actually extends the stack by. Check in stack_grow() itself
whether this has wrapped, rather than leaving it to av_extend_guts(),
which can only give a generic panic message.
Tony Cook [Wed, 28 Oct 2015 03:11:12 +0000 (14:11 +1100)]
fix the non-Win32 build breakage introduced in
ce9582af
On Unix, EU::MM was falling back to EU::MM::version::vpp, which
was removed above.
Make the real version.pm available instead.
Daniel Dragan [Tue, 27 Oct 2015 16:07:51 +0000 (12:07 -0400)]
Win32 parallel build fixes C++
-dmake's parallel scheduler is poor and has problems finding work to run
leading to idle cores, see note in commit
c2c7bda088 about
generate_uudmap.exe target so compile+link perlglob.exe in 1 process run
-remove whitespace from LIBFILES as much as possible while keeping some
prettyness in the makefile. This is so the console isn't flooded as much
with a wall of text as before.
-although perlglob.exe is very small (1 main func, that is it),
add $(OPTIMIZE), it previously wasnt CC optimized at all
-when -xc++ flag is used (USE_CPLUSPLUS=define), the command options
file is compiled like C code and syntax errors, use -x to reset file type
to nothing/auto
g++ -xc++ -I.\include -I. -I.. -DWIN32 -DPERLDLL -DPERL_CORE -s -O2 -DPERL_TEXT
MODE_SCRIPTS -DPERL_IMPLICIT_CONTEXT -DPERL_IMPLICIT_SYS -fwrapv -fno-strict-ali
asing -mms-bitfields -o..\generate_uudmap.exe ..\generate_uudmap.c -s -L"c:\per
l\lib\CORE" -L"C:\MinGW\lib" \
C:\Users\Owner\AppData\Local\Temp\mk10
C:\Users\Owner\AppData\Local\Temp\mk10:1:7: error: expected constructor, destruc
tor, or type conversion before '(' token
INPUT ( -lmoldname -lkernel32 -luser32 -lgdi32 -lwinspool -lcomdlg32 -ladvapi
32 -lshell32 -lole32 -loleaut32 -lnetapi32 -luuid -lws2_32 -lmpr -lwinmm -lvers
ion -lodbc32 -lodbccp32 -lcomctl32 )
^
dmake: Error code 129, while making '..\bitcount.h'
-since makefile.mk does not create perl523.lib during link time of
perl523.dll anymore, to allow parallelism the interface (def file and
.exp/.lib files) between libperl and XS modules is created before libperl
and XS modules are every created, this allows the 2 to build in parallel
and not be dependent on each other. This caused a link failure where an
XS module wanted a C++ mangled perl data symbol, while the def/lib file
only had the extern "C", C named data symbol (unless you take
extraordinary measures, the def/lib file is always extern "C" even if
the symbol is mangled inside C/C++ lang world and for static linking
purpose), so make all data symbols EXTERN_C, not extern, in the headers
link -out:..\..\lib\auto\PerlIO\encoding\encoding.dll -dll -nologo -nodefaultlib
-debug -opt:ref,icf -ltcg -libpath:"c:\perl\lib\CORE"
-machine:x86 "/manifestdependency:type='Win32' name='Microsoft.Windows.Common-Co
ntrols' version='6.0.0.0' processorArchitecture='*' publicKeyToken='
6595b64144cc
f1df' language='*'" -subsystem:console,"5.01" encoding.obj "..\..\lib\CORE\per
l523.lib" oldnames.lib kernel32.lib user32.lib gdi32.lib winspool.lib comdlg32.l
ib advapi32.lib shell32.lib ole32.lib oleaut32.lib netapi32.lib uuid.lib ws2_32.
lib mpr.lib winmm.lib version.lib odbc32.lib odbccp32.lib comctl32.lib msvcrt.li
b -def:encoding.def
Creating library ..\..\lib\auto\PerlIO\encoding\encoding.lib and object ..\..
\lib\auto\PerlIO\encoding\encoding.exp
encoding.obj : error LNK2001: unresolved external symbol "__declspec(dllimport)
struct _PerlIO_funcs const PerlIO_perlio" (__imp_?PerlIO_perlio@@3U_PerlIO_funcs
@@B)
..\..\lib\auto\PerlIO\encoding\encoding.dll : fatal error LNK1120: 1 unresolved
externals
dmake: Error code 224, while making '..\..\lib\auto\PerlIO\encoding\encoding.dl
l'
-------------------------------------------------------------------------
link -dll -out:..\perl523.dll -nologo -nodefaultlib -debug -opt:ref,icf -ltcg
-libpath:"c:\perl\lib\CORE" -machine:x86 "/manifestdependenc
y:type='Win32' name='Microsoft.Windows.Common-Controls' version='6.0.0.0' proces
sorArchitecture='*' publicKeyToken='
6595b64144ccf1df' language='*'" -subsystem:c
onsole,"5.01" \
@Extensions_static \
@C:\Users\Owner\AppData\Local\Temp\mk11
perl523.exp : error LNK2001: unresolved external symbol _PL_interp_size
perl523.exp : error LNK2001: unresolved external symbol _PL_interp_size_5_18_0
perl523.exp : error LNK2001: unresolved external symbol _PerlIO_pending
perl523.exp : error LNK2001: unresolved external symbol _PerlIO_perlio
..\perl523.dll : fatal error LNK1120: 4 unresolved externals
dmake: Error code 224, while making '..\perl523.dll'
With this commit, "dmake all" VC C++ build succeeds to the end, G++ build
won't build until 1 other issue is fixed, I fixed that issue for testing
but it is not in this patch, so only that issue remains preventing a G++
build from running to the end.
Daniel Dragan [Sun, 25 Oct 2015 22:53:13 +0000 (18:53 -0400)]
don't distribute version::vpp/EUMM::version::vpp
version::vpp is not part of the public API of version::, the core
cpan/version/lib/version.pm is not capable of using it unlike the
version.pm on cpan, and the Makefile.PL from cpan version:: that picks
between vpp and vxs isn't in core either. The xsubs behind core version.pm
are permanently baked in universal.c, they arent "static XS" or
"dynamic (shared lib) XS", and no XSLoader/DynaLoader is used. vpp.pm is
therefore useless in blead distributed version:: since vpp.pm is only
useful on very old perls. In blead, vpp only existed for the purpose of
making 00impl-pp.t pass, yet takes up space in the perl tarball, and was
installed into the final install location, so remove vpp.pm to save space
since it is unusable. vpp.pm and the rest of version is developed on
cpan, not in core, and vpp.pm is sort of a devel tool and sanity check and
therefore an author test, and core doesn't include author tests like
pod-coverage.t so that is one other reason it is being removed.
Also remove EUMM's vpp.pm, first it isn't miniperl compatible, and it also
is a copy more or less of version::vpp, and blead perl comes with
version.pm, since it is blead perl and not an old perl. If there is an
accident/error/bad behaviour EUMM::vpp.pm can get loaded
( http://www.nntp.perl.org/group/perl.perl5.porters/2015/10/msg232039.html
), but it is an error for it to ever be loaded. Prevent "silent failure"
by deleteing EUMM::vpp.pm, this way the failure will be an obvious
can't find EUMM::vpp.pm instead of subtle differences between the XS and
PP version implementations, or the not miniperl compatible failure.
Although ExtUtils::MakeMaker::version::regex.pm could be deleted for
the same reasons as EUMM::vpp.pm, I am leaving it in for now until the
EUMM patch in "version PP is not PP" goes through review, but still dont
install it, blead comes with the official version::regex.pm
version::vpp.pm is 22KB
00impl-pp.t is 0.5KB
ExtUtils::MakeMaker::version::vpp.pm is 23KB
ExtUtils::MakeMaker::version::regex.pm is 5KB
The 3 .pm files were being installed into the final installed perl location
where they are useless on blead perl. Some people complain perl core is
too big/bloated (redhat perl), removing 50KB and 3 files from final
location, and 45.5KB and 3 files from the tarball helps to trim the
core.
Jarkko Hietaniemi [Tue, 27 Oct 2015 12:45:28 +0000 (08:45 -0400)]
Also IRIX seems to have NaN comparison issues.
The issues seem to be much harder to trigger (harder than in VC6 or
Tru64, that is), though: the perl #125298 is the only spot where
the problem has surfaced so far. Though see also perl #126396,
with long doubles (the #125298 is with plain doubles).
All tests pass now in IRIX in blead with default config,
including the t/op/infnan.t. With -Duselongdouble nothing
new breaks (known issues with locales and M::BI)
Steve Hay [Tue, 27 Oct 2015 17:38:55 +0000 (17:38 +0000)]
Upgrade Unicode-Normalize from version 1.21 to 1.23
Steve Hay [Tue, 27 Oct 2015 17:37:24 +0000 (17:37 +0000)]
Upgrade Math-BigInt from version 1.999704 to 1.999705
Jarkko Hietaniemi [Tue, 27 Oct 2015 01:38:09 +0000 (21:38 -0400)]
Replace two ugly casts in reg_recode() calls.
With one ugly cast inside the reg_recode() call.
Karl Williamson [Fri, 25 Sep 2015 18:18:33 +0000 (12:18 -0600)]
dist/ExtUtils-CBuilder/t/02-link.t: Skip on os390
This module needs a little work for os390. But until someone
wants/needs to have it done, we'll simply skip the test.
Karl Williamson [Mon, 26 Oct 2015 17:38:53 +0000 (11:38 -0600)]
regen podcheck db
Commit
9846bace45546ab19676c8577ccf9218ddd7931d added some overlong
verbatim lines
Karl Williamson [Sun, 25 Oct 2015 02:06:10 +0000 (20:06 -0600)]
XS-APItest:fetch_pad_names.t: Comments, skip message
Karl Williamson [Sun, 25 Oct 2015 02:11:47 +0000 (20:11 -0600)]
t/re/subst.t: Don't skip a test on EBCDIC
There is no longer any reason to skip this test.
Karl Williamson [Sun, 25 Oct 2015 02:11:13 +0000 (20:11 -0600)]
re/subst.t: Clarify a test's EBCDIC skip message
Karl Williamson [Sun, 25 Oct 2015 02:09:40 +0000 (20:09 -0600)]
re/pat_rt_report.t: Clarify EBCDIC skip msg for a test
Karl Williamson [Sun, 25 Oct 2015 02:08:17 +0000 (20:08 -0600)]
t/op/print.t: Improve EBCDIC skip msg
Karl Williamson [Sun, 25 Oct 2015 02:07:51 +0000 (20:07 -0600)]
t/op/chr.t: Improve EBCDIC skip msg
Karl Williamson [Sun, 25 Oct 2015 02:07:24 +0000 (20:07 -0600)]
t/base/lex.t: Generalize for EBCDIC
Tony Cook [Mon, 26 Oct 2015 04:59:53 +0000 (15:59 +1100)]
[perl #126452] partly
0b057af7 revert for C++ builds
Daniel Dragan [Sun, 25 Oct 2015 23:57:11 +0000 (19:57 -0400)]
XS staticing in ext and dist
None of these symbols are exported on Win32 (listed in Makefile.PL with
EUMM's FUNCLIST), so they shouldn't be exported on Linux. Making them
static saves space in the SOs by removing symbol name strings, and removing
runtime plt/got indirection.
Ricardo Signes [Mon, 26 Oct 2015 01:26:37 +0000 (21:26 -0400)]
Carp: prepare for a new CPAN release
Jerry D. Hedden [Fri, 23 Oct 2015 23:55:17 +0000 (19:55 -0400)]
Upgrade to Thread::Queue 3.07
Committer: Add additional email address for contributor.
Jarkko Hietaniemi [Thu, 22 Oct 2015 12:35:10 +0000 (08:35 -0400)]
hv_iternext can return NULL
Coverity #104801
Jarkko Hietaniemi [Thu, 22 Oct 2015 11:36:56 +0000 (07:36 -0400)]
Ascertain that the fd for fcntl is not negative
Coverity id #45354
Since maxsysfd is I32 Coverity cannot prove that being larger than
it means that the fd is non-negative.
Jarkko Hietaniemi [Thu, 22 Oct 2015 11:31:58 +0000 (07:31 -0400)]
Make FITS_IN_8_BITS() always true under Coverity
Similar in spirit to
3e94db23
Coverity id #28938
Coverity id #104778
Coverity id #131329
Steve Hay [Fri, 23 Oct 2015 07:31:47 +0000 (08:31 +0100)]
Fix the (serial) build with dmake on Windows
The switch to building non-XS modules last in win32/makefile.mk (introduced
by design as part of the changes to enable parallel building) causes the
build of POSIX to break due to problems with the version module.
This change is the simplest of several workarounds/fixes put forward in
the following threads:
http://www.nntp.perl.org/group/perl.perl5.porters/2015/10/msg231903.html
http://www.nntp.perl.org/group/perl.perl5.porters/2015/10/msg232039.html
It appears that the version module could benefit from some work to render
this change unnecessary, but for now it's the least invasive way to
restore the dmake build. Thanks to all for chasing this down.
Karl Williamson [Fri, 25 Sep 2015 18:21:03 +0000 (12:21 -0600)]
lib/ExtUtils/t/Embed.t: Fix for z/OS
Karl Williamson [Wed, 21 Oct 2015 17:47:25 +0000 (11:47 -0600)]
Add test for \b{sb}
It turns out that the merge commit
f0bd363c36d925d8d3dfe3b68715763c850b171a makes \b{sb} work much better,
with the main improvement being by commit
43a7bd62098d36de61176a40d7eb3e72a2b9f033.
We were puzzled as to why it didn't seem to work as advertised, yet
passed all the extensive tests furnished by Unicode. The answer appears
to be that that those tests mostly test pair-wise things, and the bug
was because results were getting overwritten for longer inputs.
This new test is a simple, but long one.
Karl Williamson [Wed, 21 Oct 2015 17:34:47 +0000 (11:34 -0600)]
t/re/pat_advanced.t: Add comment
Karl Williamson [Wed, 21 Oct 2015 04:22:41 +0000 (22:22 -0600)]
pp_hot.c: Add comment
Karl Williamson [Wed, 21 Oct 2015 04:08:30 +0000 (22:08 -0600)]
utf8.h: Change formal macro param name to match docs
Karl Williamson [Thu, 22 Oct 2015 03:30:14 +0000 (21:30 -0600)]
PATCH: [perl #126253] Nested quantifiers not caught
Commit
4fa6dd16d2149c2aeeb32633e3a796d5ebc5b657 added a message
when a quantifier was useless, but then caused the parse to skip it, so
that if it was in an illegal combination, that was no longer caught.
Karl Williamson [Wed, 21 Oct 2015 18:45:46 +0000 (12:45 -0600)]
PATCH: [perl # 126178] Unterminated /(?i/
Chris 'BinGOs' Williams [Wed, 21 Oct 2015 17:58:58 +0000 (18:58 +0100)]
Update Time-Piece to CPAN version 1.31
[DELTA]
1.31 2015-10-20
- No Changes since 1.30_01
1.30_01 2015-09-01
- Ignore some tests on non *nix platforms
- fix compile warnings
- Inherit from Dynaloader (fix static build issues)
- Fix windows mem corruption
Chris 'BinGOs' Williams [Wed, 21 Oct 2015 17:57:07 +0000 (18:57 +0100)]
Module-CoreList is 5.
20151020 on teh CPAN
Peter Rabbitson [Wed, 21 Oct 2015 02:02:19 +0000 (22:02 -0400)]
Carp: fix test not working on older toolchain
The newly introduced _dump() utility function is a faithful
representation of what explain() does in newer Test::More versions
Peter Rabbitson [Wed, 21 Oct 2015 01:58:01 +0000 (21:58 -0400)]
Carp: fix test incorrectly assuming a recent-enough Carp.pm in @INC
James E Keenan [Wed, 21 Oct 2015 01:44:41 +0000 (21:44 -0400)]
Re-run two regen/ programs to clear up test failures in t/porting/regen.t
./perl -Ilib regen/regcharclass.pl
./perl -Ilib regen/mk_invlists.pl
Jarkko Hietaniemi [Wed, 21 Oct 2015 00:28:25 +0000 (20:28 -0400)]
Note the resource hungriness of mktables.
Daniel Dragan [Tue, 20 Oct 2015 22:03:45 +0000 (18:03 -0400)]
make non-zero exit from Makefile.PL fatal in make_ext.pl
A non-zero exit is fatal according to
http://www.nntp.perl.org/group/perl.qa/2008/08/msg11236.html so do not
continue building even if a Makefile was generated (an END block or code
after WriteMakefile() could have died for example).
This patch is from trying to fix problems in this thread
http://www.nntp.perl.org/group/perl.perl5.porters/2015/10/msg231903.html
Steve Hay [Tue, 20 Oct 2015 23:09:32 +0000 (00:09 +0100)]
Prepare Module::CoreList for 5.23.5
Steve Hay [Tue, 20 Oct 2015 22:53:22 +0000 (23:53 +0100)]
Bump version to 5.23.5
Steve Hay [Tue, 20 Oct 2015 22:42:55 +0000 (23:42 +0100)]
Create perldelta for 5.23.5
Steve Hay [Tue, 20 Oct 2015 22:30:33 +0000 (23:30 +0100)]
Tick off 5.23.4