Hey everyone,

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

Enjoy!

June 12th-19th

News

Eric Herman released Perl 5.27.1!

Tony Cook added a new Travis CI configuration, based on work by c9s and Dennis Kararsemaker. You can read more at Perl #123981.

Grant Reports

• Dave Mitchell TPF Grant 2 reports #179, #180.

Issues

New Issues

• Perl #131545: ExtUtils::MakeMaker 7.28+ MM_Win32.t fails t/porting/exec-bit.
• Perl #131559: ExtUtils::Command::MM not installing files in vendor arch-specific directories.
• Perl #131570: SIGBUS in Perl_leave_adjust_stacks().
• Perl #131571: Bleadperl v5.27.0-138-g64278e8ca7 breaks SONGMU/L-0.01.tar.gz.
• Perl #131575: s// on utf8 string occasionally crashes with "Malformed UTF-8 character".
• Perl #131577: heap-use-after-free (READ of size 1) in S_reghop4().
• Perl #131603: Wrong bracket type in perlre.pod.

Resolved Issues

• Perl #123981: [PATCH] f8210d2 Add a simple .travis.yml.
• Perl #129183: perl -S erroneously allows \ escapes in PATH.
• Perl #131336: Encode CPAN version 2.78 broke g++ build on FreeBSD-10.3.
• Perl #131487: VS, Makefile, makefile.mk.
• Perl #131522: Spurious "Assuming NOT a POSIX class" warning.
• Perl #131526: heap-buffer-overflow (READ of size 1) in Perl_my_atof2().
• Perl #131575: s// on utf8 string occasionally crashes with "Malformed UTF-8 character".
• Perl #131577: heap-use-after-free (READ of size 1) in S_reghop4().
• Perl #131594: Bleadperl v5.27.0-141-g5d09ee1cb7 breaks DAVIDO/JSON-Tiny-0.56.tar.gz.
• Perl #131603: Wrong bracket type in perlre.pod.

Patches

Tony Cook provided a patch to resolve Perl #131597: NULL pointer reference in Perl_newRV().

Discussion

Paul (LeoNerd) Evans asked (lib/_charnames.pm puts references in %^H) for a clarification about %^H which is leading to an improvement in the documentation on the topic.

In new versions you may not use a bitwise operation on characters above 0xFF, but as Graham Knop raises (Behavior of bitwise ops on unencountered wide characters), this violates a possible optimized way to determine a string or number value.

Dave Mitchell shows that open with encoding(UTF-8) could mask an OS error, and asks what we would consider correct behavior here.