Hey everyone,

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

Enjoy!

October 18th-31st

News and highlights

Aaron Crane released perl 5.25.6!

Encode 2.87 is released!

Term::ANSIColor 4.06 is released!

Sawyer X is looking for people to release development versions of perl next year, for the 5.27.x development series.

Jarkko Hietaniemi prepared (regarding slow tests) some reports of slow tests, along with explanations of how he constructed them. There is now also a new helper script available, Porting/harness-timer-report.pl.

If you're looking for a small issue to solve, Perl #116945 suggests reinstating the documentation in perlobj for $obj->foo::bar(...). Will this be your first contribution?

Dave Mitchell worked on the slowdown in split with list assignments in 5.24 raised in Perl #127999 (Slowdown in split + list assign), and provided statistics about his recent work. It's now considerably faster than before.

Yves Orton merged the patch Todd Rinaldo provided to improve the hek struct, along with a suggested assertion addition my Bulk88.

Grant reports

• Dave Mitchell TPF Grant #2 report #144, #145, #146.

Issues

New issues

• Perl #129916: CV symbol table optimization only works in main::.
• Perl #129917: perlpolicy does not explain what a "maintenance branch" is before stipulating restrictions on it.
• Perl #129918: Central documentation of dual-lived modules hard to find.
• Perl #129927: error messages refer to 'do' as 'require'.
• Perl #129928: do "a\0b" dies instead of returning undef / ENOENT.
• Perl #129950: Some UTF-8 regular expression matches fail when read from file.
• Perl #129953: lib/locale.t: Test failures and segfaulting on FreeBSD-11 and FreeBSD-12.
• Perl #129954: dist/Carp/t/arg_string.t: Test fails in blead.

Jarkko Hietaniemi noticed (newly crashing test) a test has begun to fail on OS X. Dan Collins is looking to revise it to capture the crashing in the test better.

Mathieu Arnold reported he cannot build texinfo with the latest version of perl.

Resolved issues

• Perl #6870: Odd parsing of do ... for ....
• Perl #7614: accept() cores interpreter.
• Perl #23624: Scoping of @+/@- when used with tie().
• Perl #36664: Strange behavior of shared array.
• Perl #41121: Segmentation fault Running binmode(STDOUT, ":encoding(utf8)") with ithreads.
• Perl #113934: eval 'BEGIN { goto ... }'.
• RT#129085
• Perl #128997: heap-use-after-free Perl_do_print.
• Perl #129085: heap-buffer-overflow Perl_re_intuit_start.
• Perl #129130: heap-buffer-overflow Perl_pp_chdir.
• Perl #129229: Fix Parallel Building.
• Perl #129802: Regexp performance problem with swiped strings (?).
• Perl #129895: OpenBSD 6 does not provide pid or uid in SA_SIGINFO.
• Perl #129928: do "a\0b" dies instead of returning undef / ENOENT.
• Perl #129950: Some UTF-8 regular expression matches fail when read from file.
• Perl #129954: dist/Carp/t/arg_string.t: Test fails in blead.

Suggested patches

Pali provided an updated patch for Encode.

Tony Cook provided a fix for Perl #129000 (heap-buffer-overflow Perl_utf8_hop).

Lukas Mai provided a patch for Perl #129928 (do "a\0b" dies instead of returning undef / ENOENT).

Lukas also provided a patch for Perl #116945: (Documentation on $obj->foo::bar(...) missing from 5.16 perlobj).

While working on supporting the clang -Weverything flag, Andy Lester provided a patch in Perl #129961 to simplify two small bits.

Hauke provided a patch in Perl #129967: to document @ISA in perlvar.

Discussion

There's an interesting (and tricky) regex problem raised in Perl #129903 (regexec.c stack overflow, raised by Brian Carpenter) that several people (Yves Orton, Hugo van der Sanden, Dan Collins, Father Chrysostomos) are trying to solve.

Bulk88 suggested making AUTOLOAD calling compile-time, but this would only work if the AUTOLOAD CV does not have any side-effects.

Joseph Brenner asked (perlnewmod: module-starter and ExtUtils::MakeMaker?) if we should document Module::Build::Tiny instead of Module::Starter (which creates a default with ExtUtils::MakeMaker) in perlnewmod.

Matt (mst) Trout found (Re: [perl #127810] Provide -Dfortify_inc Configure option to remove . from @INC) that an unexpected behavior to many people with do is that it loads files relative to @INC, like require. This might be relevant with the removal of . from @INC, coming up in 5.26.

Andy Lester pointed out is that if you provide undef as an the third argument in open will create an anonymous temporary file and wonders whether it should be documented properly.

Aristotle Pagaltzis observed the Unicode bug in the range operator, but he notes he is uncomfortable with fixing it immediately because of code which might inadvertently rely on it.