On Mon, Feb 07, 2022 at 08:11:55PM +0100, Alexander Bluhm wrote: > On Tue, Jan 25, 2022 at 12:26:05PM -0800, Andrew Hewus Fresh wrote: > > On Tue, Jan 25, 2022 at 09:20:55PM +0100, Alexander Bluhm wrote: > > > On Tue, Jan 25, 2022 at 12:05:48PM -0800, Andrew Hewus Fresh wrote: > > > > On Tue, Jan 25, 2022 at 06:45:12PM +0100, Alexander Bluhm wrote: > > > > > On Tue, Jan 25, 2022 at 05:13:01PM +0100, Alexander Bluhm wrote: > > > > > > On Sat, Jan 22, 2022 at 02:24:51AM +0100, Marc Espie wrote: > > > > > > > Or we can automate this with something like this: > > > > > > > > > > > <SNIP> > > > > > Our Devel::PPPort is too old. We ship with 3.57, p5-CDB_File and > > > > > p5-Moose > > > > > ship with ppport.h generated by 3.62. > > > > > > > > We could update Devel::PPPort in base. > > > > > > This will fix the current problem. But we can always have an old > > > Devel::PPPort in base and a have module in ports that comes with > > > and needs a new ppport.h. > > > > > > Somehow the porter should have a mechanism to handle this. Only 2 > > > of 300 ports that I test have issues, so it is a rare action. Per > > > default espie@'s idea works well. If we manually fix 2 ports that > > > is fine for me. > > > > > > Or we replace ppport.h only if it is outdated. > > > > We could add espie's target to those two ports. > > How do we proceed? > > I have 62 warnings in base build. This ppport was not updated yet: > /usr/src/gnu/usr.bin/perl/cpan/Term-ReadKey/ppport.h > > I have 17558 warnings in my ports test logfile. This is annoying. > Still I don't understand what this warning is all about. I looks > like llvm wants to be smarter than 30 years of preprocessor hacks. > > On CPAN is a new version 3.64 of Devel::PPPort that has fixed the issue. > https://metacpan.org/dist/Devel-PPPort/changes > https://github.com/Dual-Life/Devel-PPPort/commit/e124e16191f3c971c6efad3f9def300e2ee57ef8 > > I think we should start updating Devel::PPPort in base. Then we > can regenerate other ppport.h in base and ports. Note that some ports > need newer Devel::PPPort than we have now. So updating seems > unavoidable. > > databases/p5-CDB_File and devel/p5-Moose build with this version. > > Do we want to go this way? This seems reasonable to me.
Perl 5.34 ships with 3.62, so if that works and we do want to get in 5.34 at some point we might want to stick with that. We can also keep the newer version until we get to 5.36 which I assume comes with at least this version. I guess if we are going to auto-update the ppport.h we'll just have to keep following newer versions as they get released ahead of whatever perl we have when our updates require them? Kind of annoying, but probably the best solution. > I have untared Devel-PPPort-3.64.tar.gz into > /usr/src/gnu/usr.bin/perl/dist/Devel-PPPort and cvs added new files. > Is this the correct way to update base modules? Pretty much, yes. If it added any new files they may need to be added to the MANIFEST and new dirs to Makefile.SH (I think just for make clean). https://github.com/afresh1/OpenBSD-perl/blob/blead/patches/GOOD/put_OpenBSD-MkTemp_in_MANIFEST.patch I maintain them separately, although just Term-ReadKey from the CPAN. https://github.com/afresh1/OpenBSD-perl/tree/blead/files > Some test fail, but that is not worse than before. > -t/ppphtest.t ..... Failed 179/238 subtests > +t/ppphtest.t ..... Failed 178/235 subtests I know perl tests don't like to run in /usr/src for reasons I don't recall, but I do know that normally I can run tests with all the patches applied. > Diff has 22948 lines, so I do not send it per mail. > cvs:~bluhm/Devel-PPPort.diff > > ok? ok afresh1@ on the concept, although I didn't read it. If you can build a release with it, that's good enough for me.
