Hey everyone,

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

August 8th-14th

News and highlights

Perl 5.22.3-RC3 is now available!

Perl 5.24.1-RC3 is now available!

New patches from Todd Rinaldo for removing . from @INC. More below.

Another friendly reminder from our friendly neighborhood BinGOs to document changes in perldelta pending the release of 5.25.4.

Adding a Travis CI .travis.yml to perl might be nice but there are still problems, ML:explained#238761 by Dave Mitchell.

As Jarkko Hietaniemi fixes a problem with printf %a, Zefram opens several other issues.

A new version of Encode (2.86) is out!

Unicode has a update Proposed Update: UTR #51, Unicode Emoji (Version 4.0).

An interesting conversation on munging the optree, read more below.

Grant reports

Issues

New issues

• Perl #128865: File::Spec::VMS->canonpath() incorrect with ODS-5 style directory specs.
• Perl #128867: lib/locale.t: Test failures in threaded builds on FreeBSD 10.3.
• Perl #128874: locale gubbins produces unavoidable noise.
• Perl #128888: printf %a mishandles zero precision.
• Perl #128889: printf %a mishandles negative pseudo-precision.
• Perl #128890: printf %a rounds incorrectly.
• Perl #128893: printf %a botches 0 flag for negative values.
• Perl #128894: File::Find dont_use_nlink not set on CIFS mount for Linux.
• Perl #128899: printf %a mishandles several edge cases on long doubles.
• Perl #128901: Tests fail building Perl on vmhgfs.
• Perl #128902: MakeMaker 02-xsdynamic.t failure in AIX 5.3 with blead be93048a.
• Perl #128909: printf %a mishandles exponent-crossing rounding with long double.
• Perl #128919: limited exponent range in hex fp literal with long double.
• Perl #128926: rogue } in locale.c Perl_init_i18nl10n.
• Perl #128940: "Use of uninitialized value $fh" mentions wrong variable.

Karl Williamson reported that blead will not compile on certain configurations.

Resolved issues

• Perl #39358: sort with custom subname and prototype ($$) segfaults intermittently.
• Perl #115166: Some parts of regex engine impose I32 limit on code points.
• Perl #116867: tr/// silently fails for code points above I32_MAX.
• Perl #121151: Can't compile blead with g++ 4.8.1.
• Perl #127708: Coredump in call_sv under threads.
• Perl #128340: in-place sort incorrectly preserves element lvalue identity.
• Perl #128670: Out-of-date info on DEBUGGING in perlhacktips, perlrun.
• Perl #128719: With signatures and my vars: state variable $y masks earlier declaration.
• Perl #128843: printf %a mishandles subnormals.
• Perl #128865: File::Spec::VMS->canonpath() incorrect with ODS-5 style directory specs.
• Perl #128867: lib/locale.t: Test failures in threaded builds on FreeBSD 10.3.
• Perl #128926: rogue } in locale.c Perl_init_i18nl10n.
• Perl #128940: "Use of uninitialized value $fh" mentions wrong variable.

Rejected issues

• Perl #123184: pos++.

Suggested patches

Todd Rinaldo provided new patches in Perl #127810 to remove . from @INC.

Peter Rabbitson (ribasushi) wrote an alternative patch for base.pm. Kent Fredric sent it to the list. The releases of 5.23.3 and 5.24.1 will be delayed until this patch is discussed and decided on.

Dave Mitchell provided a new patch for adding travis.yml file in Perl #123981.

Tony Cook provided a patch for fixing Perl #128263 (Assert fail in Perl_magic_get without other symptoms).

Leon Timmermans provided a patch for Perl #128902 (MakeMaker 02-xsdynamic.t failure in AIX 5.3 with blead be93048a) here.

Smylers provided a merged patch in Perl #128910 to remove state (@a) upcoming feature claim from perldiag.

Bulk88 provided a merged patch in Perl #128929 to silence MSVC warnings for NATIVE_UTF8_TO_I8/I8_TO_NATIVE_UTF8.

Tony Cook provided a patched for Perl #128095: (Failed test dist/IO/t/cachepropagate-unix on OS X with Perl 5.22, regression from 5.20).

Bulk88 provided a patch to automatically detect GCC vs VC and set VC version number on Win32 in Perl #128930.

Discussion

Jarkko Hietaniemi asked if make_ext.pl script is required every single time make perl is run. It's costly. Craig A. Berry provided some history and Nicholas Clark (who introduced this change chimed in with further explanations.

The discussion (Internals:: undocumented) around documenting internal functions continued. It resulted in Yves Orton committing changes to both move some functions and document them.

Father Chrysostomos noticed that Encode uses C++ comments instead of C comments. Perl uses C89 and guarantees support for compilers that do not understand // comments.

A suggestion on how to print Size_t and SSize_t by John P. Linderman, along with a fun anecdote about Awk, ANSI, and floats.

Michael Schroeder pinged the topic of a suggested change to base.pm. Father Chrysostomos responded with the possible problems this suggested change has.

On a fascinating thread (Flailing randomly at ops), Paul Evans (LeoNerd) asked for help understanding munging the optree to reach a particular goal he has. Father Chrysostomos provided insightful comments, and Paul is writing about it on his blog.