Hey everyone,

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

April 5th-13th

Update: It was stated that Perl #121734 was resolved, but it wasn't. This was amended below. Thank you, Tony, for the correction!

News and updates

Ricardo Signes released Perl 5.24.0-RC1!

Steve Hay released Perl 5.22.2-RC1!

Encode 2.84 has been released! You can read more here.

In preparation of Perl 5.24.0, Ricardo Signes bumped the version to 5.24.0 RC0, in an effort to shake off bugs relating to hard-coded versions. But by now, as you can tell, 5.24.0-RC1 is released!

Ricardo Signes added some files to dissuade users from filing GH pull requests. A conversation ensued on whether there's a way to get this right and whether having GH pull request functionality is worth it, and if so, how.

Issues

New issues

• Perl #127849: SKIP/TODO timing tests on VMs.
• Perl #127875: Bleadperl breaks Scope::Upper.
• Perl #127877: Concat with uninitialized variables.
• Perl #127887: printf uses wrong cached number.
• Perl #127894: -DDEBUGGING -Dusequadmath -Dusethreads builds crash early.
• Perl #127895: ext/POSIX/t/math.t fails on FreeBSD 9.1 i386.

Resolved issues

• Perl #5907: Inconsistencies in locale support.
• Perl #122551: Should work with utf-8 characters.
• Perl #123861: \C yields check_locale_boundary_crossing assertion failure.
• Perl #126162: stat() behaves differently when passed an array in argument list.
• Perl #126545: Bleadperl breaks perl5i.
• Perl #127746: charset.t and subst.t fail on Solaris under -Duse64bitall.
• Perl #127708: Coredump in call_sv under threads.
• Perl #127799: Bleadperl breaks Module::Build::Pluggable.
• Perl #127829: Bleadperl breaks Pod::Usage::Return.
• Perl #127855: Slowdown in m//g on COW strings of certain lengths.
• Perl #127882: Fix some slightly mangled documentation.

Proposed patches

Ed Avis provides an updated patch for Perl #126239.

Craig A. Berry provides a patch in Perl #127852 to finish mathomizing Perl_instr. It's meant to fix a bug which was introduced in the last month before 5.24.0 will be released. It was decided to revert meanwhile and have this patch added past 5.24.0.

Tony Cook provided a patch in Perl #127894 to prevent building with both -Duselongdouble and -Dusequadmath, allowing only one or the other.

Jim Cromie provided a patch in Perl #127880 to fix builds with -DPERL_TRACE_OPS.

Jim also provided a patch in Perl #127885 to enhance bench.pl to test same perl under different options and arguments.

Todd Rinaldo has resubmitted his patches for introducing a compile-time option that removes . from @INC.

Tony Cook provided a new patch in Perl #127780.

Discussion

Dave Mitchell proposes that we introduce the -DPERL_OP_PARENT compiler flag by default in Perl 5.25.0. This will provide room for optimizations, both in core and on XS CPAN modules. Dave explains more in the post. So far exciting responses.

Stemming from a conversation around Perl #127552 Dave Mitchell suggests warning about punctuation variables that aren't assigned a purpose yet, such as @$ which are likely to be typos.

Dave Mitchell provides a quick refresher in Perl #127663 about the safety of -i.

I liked this comment from Father Chrysostomos on Perl #8979 because sometimes an issue is fixed regardless of any issue open about it. :)

Dave Mitchell says that because of the code freeze, the patch for standardizing memory types in the core will have to wait. Andy Dougherty provides interesting details on the history of the various types.

Dave Mitchell suggested updating the documentation about reporting a problem privately. With the help of Hugo van der Sanden, revised documentation was introduced.

Aristotle Pagaltzis began a sub-thread specifically about Scope::Upper which includes comments from fellow core developers about the responsibility of p5p to support highly-used modules.

Ricardo Signes adds a detailed analysis of this problem and possible solutions.

Ricardo Signes raises the issue with the IO distribution which had no stable CPAN release since v1.25, May 2009. Todd Rinaldo is looking into the issue. From this, Aristotle Pagaltzis opened the topic of "Un-dual-life'ing" and its affect on CPAN dependencies.