Perl 5 Porters Mailing List Summary: July 27th-31st
Hey everyone,
Following is the p5p (Perl 5 Porters) mailing list summary for the past week. Enjoy!
July 27th-31st
News and updates
Now that CVE-2016-1238 has been addressed in patches for 5.24.1 and
5.22.3, Todd Rinaldo has
renewed
the discussion around removing the current directory (.
)
from @INC
entirely in core.
Leon Timmermans intends (The great pl2bat unification) to reunite three forks of the same code for writing Batch scripts that run Perl scripts.
Grant reports
Tony Cook provided (TONYC TPF Grant 7 report 15) several (TONYC TPF Grant 7 report 16) grant (TONYC TPF Grant 7 report 17) reports (TONYC TPF Grant - June 2016). I recommend reading them as they provide interesting details on the work Tony has done in core recently.
Issues
New issues
- Perl #128740: perl5 segfaults (perl5 versions gentoo(5.20.0, 5.22.0), guix (5.22.1)) without message.
- Perl #128743: Here-doc missing terminator message is not null-clean.
- Perl #128747:
base/lex
test 41 failure undermake minitest
in some platforms. - Perl #128753: Bug
#128747 for perl5:
base/lex
test 41 failure undermake minitest
in some platforms. - Perl #128763:
NV_PAYLOAD_SIZEOF_ASSERT
fails on PPC darwin. - Perl #128765:
Custom warnings categories are disabled by any preceding
no warnings 'category'
. - Perl #128769: Bleadperl v5.25.3-22-g8901dde breaks FLORA/Class-C3-XS-0.13.tar.gz.
- Perl #128774:
Signal handlers aren't honoured at
END
time. - Perl #128786:
making
$^V
a version object broke functionality.
Resolved issues
- Perl #124115: [PATCH] Present utf8 and bytes pragmas consistently in perlport.
- Perl #128574: Some tests call fresh_perl* incorrectly.
- Perl #128685: Repeated installs broken in blead.
- Perl #128701: Unicode handling of warnings appears broken in 5.18.2 and 5.24.0.
- Perl #128734:
tr///
with named Unicode chars broken since 5.23.3. - Perl #128747:
base/lex
test 41 failure undermake minitest
in some platforms. - Perl #128753: Bug
#128747 for perl5:
base/lex
test 41 failure undermake minitest
in some platforms. - Perl #128763:
NV_PAYLOAD_SIZEOF_ASSERT
fails on PPC darwin.
Rejected issues
- Perl #75506:
perldb
"m" has problems with PPI::Document objects. - Perl #128767: File::Path::make_path dies under error conditions.
The issue (Testing Config::General consumes extreme amounts of memory) raised by A. Sinan Unur has been found by Sinan to be a problem with Config::General instead.
Suggested patches
Karen Etheridge provided a patch in
Perl #128775 to
update
Parse::CPAN::Meta to
ignore PERL_YAML_BACKEND
and PERL_JSON_BACKEND
while building core
itself, where other backends are not installed yet. This patch will be
merged and added to 5.24.1.
Lukas Mai provided a patch in
Perl #128782
(perlport
hard to read with platform names following function
description) to update perlport
to make it more readable.
Father Chrysostomos provided a patch to Perl #128740: perl5 segfaults (perl5 versions gentoo(5.20.0, 5.22.0), guix (5.22.1)) without message.
Discussion
Matt Trout (mst)
suggested
to add a patch to remove .
from @INC
which will first be set to
false, then turned on by testers to try and find issues, and then fully
turned on once issues are resolved. He notes specifically that some
portions of the toolchain is the main concern of breakage.
Karl Williamson asked (Use of global vs interpreter variables) on the issue of global variables in the interpreter. Should we minimize them or add a few to cut down on memory consumption and generation?
Father Chrysostomos
added
more information on
Perl #128708
(recommended my sub
example in perlsub
leaks memory) and the
problem with lexical subroutines and memory leaks.
H. Merijn Brand (Tux)
mentioned
(versiononly commit) that a commit to update Configure
but requires
amending the file Porting/Glossary
.
Ricardo Signes updated (perl5.24.0 RC deletion) that he will delete five old RCs for 5.24.0.
Relating to Perl #128774 (Signal handlers aren't honoured at END time), Zefram explained the reason for the change and that the unhooking of signals should probably happen either before global destruction upon process closing or right after. Leon Timmermans favors before.
Father Chrysostomos added (Two maint blockers) two tickets to include as blockers for 5.24.1. The first merged, the second will be merged as well.
Jarkko Hietaniemi
worked
on adding support for clang's -Weverything
, but
mentions it does not fully work on blead yet. Andy Lester stepped
up.
Dave Mitchell is continuing (Re: merging optimising sub signature work so far) his progress on the subroutine signature optimization work.
There are additional comments (RE: ASCII support in z/OS) on the suggested change for perl on z/OS.
Aristotle Pagaltzis provided comments on Karl Williamson's RFC on syntax for allowing script run pattern matching.
Leave a comment