Hey everyone,

Following is the p5p (Perl 5 Porters) mailing list summary for the past week. Enjoy!

May 10th-18th

News and updates

Sawyer X reports that this month will see a release of another development release (instead of next month) in order to accommodate changes which waited until 5.24.0 and 5.25.0 were released.

Tony Cook provided his grant reports. In total about 32 hours and approximately 16 tickets were reviewed or working on, and 1 patch applied.

Dave Mitchell provided his grant report. The majority of the work was done to try and make Scope::Upper work on 5.24.0.

Craig A. Berry announced that binary kits of Perl 5.24.0 for OpenVMS are now available.

Issues

New issues

• Perl #128107: ExtUtils/typemap references unresolved symbol.
• Perl #128111: gettimeofday.t fails "time left should be zero".
• Perl #128139: BBC (Blead Breaks CPAN) on a long list of modules.
• Perl #128143: Inconsistent behaviour when decoding in substitution.
• Perl #128167: Issues with the embedded y2038 library.
• Perl #128170: Assert failure in regcomp.c.
• Perl #128179: Bleadperl breaks Array::Base.
• Warnings from hv_func.h.
• Dan Collins reports GCC 6.1 on Debian build report having a few warnings.

Resolved issues

• RT #100183: Carp won't print $. if it's in "chunk" mode.
• Perl #125833: require ::foo will try to load /foo.pm.
• Perl #127000: Update to latest Test-Simple dev release.
• Perl #127234: Fix the Configure escape with usecrosscompile but no targethost.
• Perl #127780: Documentation needed: backticks, qx() return octets, not characters.
• Perl #127821: lround() is not exported from POSIX.
• Perl #127852: Finish mathomizing Perl_instr.
• Perl #127880: -DPERL_TRACE_OPS builds fail make test for all other config options.
• Perl #127952: Assertion failure.
• Perl #127976: each($scalar) error message anomalies.
• Perl #128012: Support mandoc in perl5db.t.
• Perl #128020: perlbug needs to wrap long lines for MTA safety.
• Perl #128085: SIGSEGV in S_regmatch with S_study_chunk.
• Perl #128086: SIGSEGV in hek_eq_pvn_flags.
• Perl #128089: Fix Module::CoreList is_core bounds checking for specific module versions.
• Perl #128105: Clarify description of sprintf %.1g in perlfunc.
• Perl #128106: SIGSEGV in Perl_sv_resetpvn when reset catches a sub.
• Perl #128109: Issues with (?0) being wrong in 5.24 (RC?).
• Perl #128126: Fix link to Crosby paper on Algorithmic Complexity Attacks.
• Perl #128131: gcc 6 link-time optimization (LTO) breaks Configure symbol detection.
• Perl #128171: Assert fail in toke.c.

Proposed patches

Maxwell Hadyn provided a patch in Perl #128105 to clarify the description of sprintf %.1g in perlfunc.

Jim Cromie provided a patch in Perl #128112 to improve the bug handling of glibc i-modulo.

Chad Granum (Exordist) provided a patch in Perl #128113 to upgrade Test-Simple in perl blead.

Lukas Mai provided a patch for Perl #128131 to fix the link-time optimization (LTO) for GCC 6.

Salvador Fandiño provided a patch to fix a PerlIO::encoding test.

Tony Cook provided a patch in Perl #126228.

Karen Etheridge provided a patch in Perl #128153 to upgrade Module::Metadata to 1.000032 and another in Perl #128160 to silence some diagnostic messages that were printed with the first patch. Karen also submitted a patch in Perl #128169 to remove internal test modules from Module::CoreList.

Jerry D. Hedden provided patches to update threads and threads::shared to 2.08 and 1.52, respectively.

Discussion

Mojca Miklavec has taken over maintainership of perl5 in MacPorts and asks for advise on the appropriate configuration flags. An answer provided by Craig A. Berry.

James E. Keenan set up a smoke branch to test the upgrade patch in Perl #128113 (mentioned above) and asks whether we should still merge to blead for sensitive distributions. Tony Cook intends to apply the patch soon. Tony provided some benchmark results, and Chad Granum provided an explanation of possible differences.

The discussion about removing . from @INC is continuing, thanks to constant bumps from Todd Rinaldo. I recommend reading about the topic and its possible effects and various offered solutions.

Ricardo Signes asks if anything is preventing us from moving lexical subroutines from "experimental". Father Chrysostomos mentions Perl #123367.

H. Merijn Brand (Tux) reports he benchmarked Perl 5.24.0 and sees an improvement in speed.

Timothy Madden asked how to use I/O from native C extensions. Leon Timmermands provides an answer.

Karl Williamson reminds everyone to update all RT tickets for 5.26 pending release.

Aaron Crane asks about a proposal to deprecated and remove a C-specific function from POSIX.pm.

Following no objections, Dave Mitchell had set -DPERL_PARENT_OP as default.

Alberto Simões asks about the return keyword behavior in a map block.

Note-worthy, Aaron Crane's explanation for a particular bug.

H. Merijn (Tux) Brand had# a problem which Zefram pinned to a check done on the shebang for Perl 6 which created a false positive. There's an active discussion on whether the check should be reverted or kept.

Karl Williamson asks about Perl #57512 (implicit close()s are silently unchecked for error). Ricardo Signes explains the problem and Father Chrysostomos elaborates. It would seem that under the new context stack Dave Mitchell has written, this class of errors could be fixed.

Karl Williamson simplified and generalized mathom.c, which is a file that keeps several definitions, originally meant for binary compatibility. Dave Mitchell suggests cleaning up up after every stable release.