On Fri, 18 Dec 2020 at 11:47, Steve Hay <steve.m.hay@googlemail.com> wrote: > Apologies for this coming a month or two later than I intended, but I've > finally got the ball rolling on making 5.32.1. > > I've pushed an initial batch of proposals for backporting here: > > https://raw.githubusercontent.com/Perl/perl5/maint-votes/votes-5.32.xml > > This includes all the commits in the "5.32.1" milestone or labelled as or > "backport-5.32" plus some more. The changes from PR#18407 will be included > as well, once that is tested okay and put into blead. > > If you have any more suggestions for commits that meet the backport > criteria set out in perlpolicy.pod then please let me know. > > Committers: Please consider adding votes on the proposals so we can get > them merged into maint-5.32. > > Hopefully we can get a release out in January. > I was hoping to get an RC1 out this weekend, with the final release hopefully a fortnight later, but I'm still chasing quite a few votes: GH#16529 - win32.c: rework the waitpid(-1, WNOHANG) fix GH#18341 - Avoid deadlock with PERL_MEM_LOG GH#18380 - skip trying to constant fold an incomplete op tree GH#17816 - list assign in list context was over-optimising GH#16685 - list assign in list context: honour LHS undef GH# 18063 - die_unwind(): global destruction GH#18406 - Make DynaLoader on MacOS check library existence with dlopen 0287c5eaf3 - Make MakeMaker use dlopen to check for library availability on MacOS GH#18408 - Correctly set LC_CTYPE back in POSIX/t/time.t GH#18097 - Make security warnings in the Safe and Opcode modules more direct Also, it has been suggested that the proposed cherry-pick of 31667aca63 (Fix cygwin build) depends on 7953f73fd8 (Name individual locale locks). If so then I will drop 31667aca63, but it would be great if somebody could do a test build of the current maint-5.32 branch on Cygwin to see whether it's currently in need of a "Fix cygwin build" change. Finally, the group of commits that fix GH#18123 have been voted on, but I'm unsure whether 55afc78396 should really be backported or not because it changes an error message. perlpolicy.pod says that we should not backport commits that add new warnings or errors. Is it acceptable to change the wording of an existing warning or error in a maint release? The commit in question changes the fatal error "Your regular expression used C<\K> in a lookhead or lookbehind assertion, which isn't permitted." to "Your regular expression used C<\K> in a lookhead or lookbehind assertion, which currently isn't permitted." It's such a minor change that I feel it's acceptable, but I want to double-check before going ahead in case of any strong objections.Thread Previous | Thread Next