Hey everyone,

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

June 22nd-29th

News and updates

Steve Hay has created a ticket to monitor all the release blockers for perl 5.22.3, Perl #128491.

Issues

New issues

• Perl #128450: __SUB__ in format behaves inconsistently.
• Perl #128451: Win32: $Config{libpth} is broken in 5.25.2.
• Perl #128465: set_mro too lazy, permits unmergeable inheritance.
• Perl #128466: perl debugger magic comparison.
• Perl #128479: Assigning hashref to symbol table not an error.
• Perl #128482: Bleadperl breaks Tie::REHash.
• Perl #128484: Configure: tweak config.sh for easier noise-free checksum.
• Perl #128487: Win32: Signaling NaN is evasive on recent 32-bit perls.

Resolved issues

• Perl #127585: IO-Compress test hangs on MS Windows.
• Perl #128238: Assert fail in gv.c without other symptoms.
• Perl #128252: Assert failure in op.c without other symptoms.
• Perl #128464: Sync CPAN Locale::Maketext 1.27 with blead.
• Perl #128478: Change in behaviour of $foo::$bar.

Discussion

Yves Orton has changed the sort order of the MANIFEST file.

Yves Orton also introduced his change of the return signature of scalar(%hash) to match 0+keys(%hash) as a continuation of Perl #114576. In short, it didn't work the way you thought, and provided details that were not helpful for you to begin with. :)

David Farrell asked about a large amount of stat calls he was seeing from perl, and wondered if they were all necessary or whether it was possible to reduce them. Some options raised: Compiling Perl without .pmc support, and reducing the number of directories in @INC. Zefram notes the purpose of continuing to iterate through directories despite not finding a module in them. Additionally with Linux detry caches, the actual stat calls would be faster than observed.

However, at the end, Dave Mitchell showed an indistinguishable difference for the stat calls between 2 @INC entries and 8.

Unicode provides a Script_Extensions property (scx) which Karl Williamson suggests using as the underlying for the single-value synonym in Perl (instead of the currently-used Script property, sc).

Darren Duncan asked if it were possible to release perl 5.24.1 with List::Util 1.45, since it now carries a stable implementation of uniq. This will likely not happen because maintenance releases should not include new features, and James E. Keenan quotes the paragraph from perlpolicy.

John Siracusa reports some confusion around eval and $@, which seems to be a bug.

Marc Lehmann has been banned from p5p for violating the Standards for Conduct.