-=head1 TODO
-
-The following changes are a filtered list of changes which weren't backported
-to 5.10.1. They were run through changelogger.bestpractical.com and triaged
-by a set of Perl 5 Porters. Changes to blead after 5945e41e have not yet
-been triaged or integrated.
-
-The following changes need to be
-
-1) deleted if they don't merit inclusion
-
- OR
-
-2) lightly copyedited and integrated into the perldelta above. Unfortunately, many of our
-commit messages are somewhat terse and require a bit more help to turn into something readable
-
-
-
-
-=head1 API
-
-=head2 Move the reg_stringify logic to Perl_sv_2pv_flags
-
-=head2 mg_copy ought to take an I32
-
-=head2 Perl_store_cop_label() isn't meant to be part of the public API.
-
-=head2 Perl_gv_fetchmethod{,_autoload,_flags} are actually never* called with a non-NULL stash.
-So change the parameter to NN.
-
-
-=head2 Promote Perl_setdefout() to the public API.
-
-=head2 Add get_cvs() as a shortcut for STR_WITH_LEN() and Perl_get_cvn_flags()
-
-=head2 In Perl_newCONSTSUB(), sv should not be NULL.
-
-=head2 GvUNIQUE* have been defined as 0 since 2005/06/30 - high time to remove them.
-
-=head2 invert and rename PERL_MEM_LOG_STDERR to PERL_MEM_LOG_NOIMPL
-
-Most users who want PERL_MEM_LOG want the default implementation,
-give it to them. Users providing their own implementation can
-obtain current behavior by adding -DPERL_MEM_LOG_NOIMPL.
-Frankly, the average user probably wants _ENV by default too.
-
-=head2 simplify PERL_MEM_LOG
-
-This combines multiple environment variable reads into 1,
-where it looks for values like "2mst"
--2 leading digits are atoi()d to get FD
--m memory logging please
--s sv logging also
--t timestamp those please.
-
-Combining these reduces overhead such that it seemed
-worthwhile to drop all the ifdefs. TBD whether this works
-in the environment that drove the original tradeoffs.
-
-If it isnt enough, Id be tempted by a global static ptr,
-and on 1st use, is read, seen 0, a lock is taken, and getenvar
-run to populate it, unlocked, proceed. This would remove
-iterative overheads.
-
-=head2 Add a parameter "destructing" to Gv_AMupdate()
-
-This boolean parameter indicates if the function has been called
-to update the overload magic table while looking up the DESTROY
-method. In this case, it's probably best to avoid croaking if
-those tables could not be updated (for example due to a method
-that could not be loaded.)
-
-=head2 Modify the return value of Gv_AMupdate to indicate a compilation error
-
-This way we'll restore most of the performance on object desctruction
-lost by the previous commit
-
-
-=head2 local $SIG{FOO} = sub {...}; sets signal handler to SIG_DFL
-
-Re: [perl #60360] [PATCH] UPDATED: local $SIG{FOO} = sub {...}; sets signal handler to SIG_DFL
-Message-ID: <20081112234504.GI2062@tytlal.topaz.cx>
-
-Updated patch to retain source compatibility.
-
-Plus using the correct PERL_ARGS_ASSERT_SAVE_HELEM_FLAGS
-macro and running make regen.
-
-=head2 Respecting inc_version_list while processing PERL_VENDORLIB_STEM
-
-Respecting inc_version_list while processing PERL_VENDORLIB_STEM
-From: "Mandalemula, Rajesh" <Rajesh.Mandalemula@deshaw.com>
-
-=head2 Change PL_debug behaviour
-
-
-String eval lines are now saved whenever
-a subroutine is defined, even if the eval'd string has subsequent
-syntax errors. This allows the debugger to single step into these
-subroutines.
-
-
-=head2 Tied filehandles now have an additional method EOF which provides the EOF type
-
-=head2 Perl is now smarter about adding a -I dir to the beginning or end of @INC
-
-=head2 On scope end, delete localized array elements that should not exist anymore, so that the array recovers its previous length. Honour EXISTS and DELETE for tied arrays.
-
-=head2 When a glob is deleted, mark its sub as ANON.
-
-=head2 Require a space or a newline after a "#line XXX" directive
-
-=head2 Forbid using "foreach" as an attribute
-
-(like all other control flow statements)
-
-
-=head2 Unregister signal handlers before destroying my_perl
-
-If the signal handler runs after perl_destruct() has been called, it
-will get an invalid (or NULL) my_perl when it asks for the
-thread-specific interpreter struct. This patch resets the signal
-handler for any signal previously handled by PL_csighandlerp to SIG_DFL
-before calling perl_destruct().
-
-=head2 Perl_magic_clearsig() needs to remove magic, else delete $SIG{INT} returns undef
-
-Perl_magic_clearsig() needs to remove magic, else delete $SIG{INT} returns undef
-instead of the now-removed INT handler.
-
-=head2 [perl #66452] TMPDIR not honored when opening an anonymous temporary file
-
-[perl #66452] TMPDIR not honored when opening an anonymous temporary file
-
-=head2 The attached patch to perlio.c fixes the problem of errno getting set.
-
-While I am firmly in the school of "do not look at $! except immediately
-after a failure", I also agree that spuriously setting it is messy. But
-there is just no way of knowing where your errno might have been.
-
-The problem was that PerlIO_fast_gets() (and other nearby similar
-capability-checking PerlIO routines) set the errno (and it was being
-called a lot, from sv_gets()). I think setting the errno here was
-a mistake: checking for "can has FOO" should not set external state,
-such as the errno. The patch removes that errno trashing from all those
-routines.
-
-=head2 Trim all trailing / from "." in @INC when filling %INC
-
-This fixes bug #66942 : as a / was left in the directory name,
-$INC{"Foo.pm"} for a file loaded from the current directory
-was given the incorrect value "/Foo.pm".
-
-=head2 Don't enqueue pending signals during global destruction
-
-Global destruction is not signal-safe. PL_psig_pend may already
-be gone when the signal handler is called (with destruct_level > 0).
-NULL it before freeing it to prevent a race condition.
-
-=head2 Eliminate struct regexp_allocated and xpvio_allocated.
-
-Calculate memory allocation using regexp and XPVIO, and the offset of the first
-real structure member. This avoids tripping over alignment differences between
-X* and x*_allocated, because x*_allocated doesn't have a double in it.
-
-
-=head1 internals
-
-=head2 [perl #47047] Use of inherited AUTOLOAD for non-method is deprecated
-
-=head2 Remove the definitions of Null(), Nullch, Nullfp, Nullsv and PL_na when code is within the perl source tree
-
-=head2 Replace our assert-which-can-be-caught-by-eval with the real deal from the standard C library.
-
-=head2 Tweak Perl_sv_upgrade() so that references can upgrade to SVt_PV
-
-=head2 Eliminate prelen from struct regexp.
-
-=head2 Change Perl_av_iter_p() to return IV* rather than I32* (which means
-
-=head2 Reorder the external regexp flags to get RXf_PMf_STD_PMMOD into the
-
-lowest 4 bits (which saves a shift), and the "flags indicating special
-patterns" into contiguous bits. This makes everything a little tidier,
-and saves 88 bytes (woohoo!) of object file with -Os on x86 FreeBSD.
-
-
-=head2 Re-implement the SvOOK() hack to store the offset as a BER encoded number in the part of the PVX that is being released.
-(It will always
-fit, as chopping off 1 byte gives just enough space for recording a
-delta of up to 127). This allows SvOOK() to co-exist with SvIOK_on(),
-which means all the calls to SvOOK_off() [with the possibility of a
-call to sv_backoff()] in SvIOK_on() can be removed. This ought to make
-a lot of straight line code a little bit simpler.
-OOK()d scalars can now be SVt_PV, as the IVX isn't needed.
-
-=head2 Abolish wraplen from struct regexp. We're already storing it in SvCUR.
-
-=head2 Make Perl_pregcomp() use SvUTF8() of the pattern, rather than the flag bit in pmflags, to decide whether the pattern is UTF-8.
-
-=head2 Abolish RXf_UTF8. Store the UTF-8-ness of the pattern with SvUTF8().
-
-=head2 In struct regexp move the member paren_names to the IV union.
-
-=head2 Make REGEXP a type distinct from SV. (Much like AV, CV, GV, HV).
-
-=head2 Allow sv_setsv_flags() to copy SVt_REGEXP much like it copies SVt_FORMAT - the just string buffer.
-
-
-=head2 Correct a long-standing ithreads reference counting anonamly
-
-The reference count only needs "doubling" when the scalar is pushed onto
-PL_regex_padav for the second time.
-
-
-=head2 In PL_regexp_padav, store regexps via real references, rather than hiding them within IVs.
-
-We can do this now that they are real SV pointers.
-
-=head2 With regexps stored as real RVs, we can eliminate SvREPADTMP().
-
-=head2 REGEXPs are now stored directly in PL_regex_padav, rather than indirectly via RVs.
-
-=head2 Remove code that protected pp_qr against REGEXPs going away during global destruction whilst they were stored via true references in PL_regex_padav.
-
-=head2 Remove PM_GETRE_SAFE and PM_SETRE_SAFE as nothing uses them.
-
-=head2 Note the U8 sized space created by removing -P, and check that it is now an illegal command line flag.
-
-=head2 Pack the recycled pad offsets into an SV at PL_regex_pad[0].
-
-=head2 Re-order so that the !SvOK() case is last (which should be rare)
-
-=head2 Extend PUSHFORMAT() to take a second parameter to set retop, to save NULLing it and then reassigning.
-
-=head2 Split struct block_sub into struct block_sub and struct block_format.
-
-Split struct block_sub into struct block_sub and struct block_format.
-(CXt_SUB and CXt_FORMAT were using some comon members, but some members
-were only for one or the other.)
-
-=head2 Change the wantarray result from caller from IV to bool for the SCALAR/ARRAY case.
-
-This doesn't contradict the documentation, as there isn't any. Oops.
-
-
-=head2 Give G_VOID, G_SCALAR and G_ARRAY the same numeric values as OPf_WANT_VOID, OPf_WANT_SCALAR and OPf_WANT_LIST.
-
-
-=head2 Squeeze the context type down to 4 bits, and move the private flags to fit within the next 4 bits.
-
-
-=head2 In struct block change blku_type from U8 to U16, and the "spare" U8 to U16, with the lockstep changes in struct subst.
-Eliminate lval from
-struct block_sub, and instead store it in the U16 in struct block.
-
-
-=head2 In struct block_eval, eliminate old_in_eval and old_op_type by storing the data in blk_u16.
-
-
-=head2 The layout for struct block_loop under ithreads can be simplified.
-
-Instead of wedging the pad offset into a void* iterdata, and always
-storing PL_comppad even when it isn't used, instead do this:
-
-PAD *oldcomppad; /* Also used for the GV, if targoffset is 0 */
-/* This is also accesible via cx->blk_loop.my_op->op_targ */
-PADOFFSET targoffset;
-
-and store the GV pointer in oldcompad. Pointers to pointers seems
-cleaner. This also allows us to eliminate the flag bit CXp_PADVAR.
-
-
-=head2 In XS_PerlIO_get_layers() take advantage of the implementation of
-
-In XS_PerlIO_get_layers() take advantage of the implementation of
-PerlIO_get_layers(), by co-opting the new SVs it creates, rather than
-copying them.
-
-
-=head2 Micro-optimise the order of the context types. [Because I can :-)]
-
-=head2 [patch] optimize OP_IS_(FILETEST|SOCKET) macros
-
-=head2 Eliminate ck_lengthconst.
-
-=head2 Chainsaw DEBUG_S out, as suggested by Vincent Pit.
-
-=head2 Unsupported private API functions are now declared "static" to prevent leakage to the public API
-
-=head2 Perl_cv_ckproto() is not part of the public API, and not used anywhere. It has been removed
-
-=head2 Remove all the 5005threads specific mutex macros, which are now vestigial.
-
-=head2 Do not honor TMPDIR for anonymous temporary files when tainting
-
-Use a default of /tmp on Unixes when TMPDIR is unset or empty, or
-when creation of a temporary file in it fails
-
-=head2 Add a pluggable hook in op_free()
-
-
-
-
-
https://perl5.git.perl.org / perl5.git / blobdiff