perl 5.36.0

[cho-m]

• Have you followed the guidelines for contributing?
• Have you ensured that your commits follow the commit style guide?
• Have you checked that there aren't other open pull requests for the same formula update/change?
• Have you built your formula locally with brew install --build-from-source <formula>, where <formula> is the name of the formula you're submitting?
• Is your test running fine brew test <formula>, where <formula> is the name of the formula you're submitting?
• Does your build pass brew audit --strict <formula> (after doing brew install --build-from-source <formula>)? If this is a new formula, does it pass brew audit --new <formula>?

---

Starting over on #102549 since some changes have gone in other PRs like #108663.

Planning to cherry-pick the commits from other PR once we have readout of linkage failures.

[danielnachun]

Perl 5.37 is out now, so we should try that instead.

[danielnachun]

Also from prior testing, the self-hosted runner is not needed just for dependent testing, so you might be able to run this faster if you remove that flag.

[cho-m]

Perl 5.37 is out now, so we should try that instead.

Odd numbers are development versions. Latest stable is still 5.36.

[danielnachun]

Perl 5.37 is out now, so we should try that instead.

Odd numbers are development versions. Latest stable is still 5.36.

Ah I did not know that. Should we add a comment explaining that?

[cho-m]

Just picked the first test from directory. Not sure if there is a better one. We've been hit a few times by upstream modifying the test.bcf file with new releases. Unless we save a copy, it didn't seem worth seeing CI failures related to checksum mismatches or version incompatibility.

[danielnachun]

I didn't realize there was test directory but that seems like a much more reliable solution.

[danielnachun]

Linux-specific failures:

• subversion: top-level mandir
• feedgnuplot: top-level mandir
• grokj2k: we're on a very old release - the newest release builds fine on Linux
• pidgin: needs a fix for XML::Parser - it should use ENV.prepend_path instead of ENV.prepend when adding the library path of intltool to PERL5LIB.

[cho-m]

• grokj2k: we're on a very old release - the newest release builds fine on Linux

I think we've had issues on macOS when trying to update grokj2k. Right after I fixed formula in #95683 (9.2.0 --> 9.7.1), upstream broke build in next patch #96168 and we haven't been able to update since.

Not sure about current version, but it was one of failures in previous perl PR #102549 so I want to avoid blocking perl update on this.

---

• pidgin: needs a fix for XML::Parser - it should use ENV.prepend_path instead of ENV.prepend when adding the library path of intltool to PERL5LIB.

I was previously wondering if we should enable perl vendor directory and create a perl-xml-parser formula just to avoid all these workarounds due to common intltool.m4 script.

Though, most formulae are moving away from intltool so probably not important.

[cho-m]

Looks like abricate is failing everywhere. Given we don't have a Monterey bottle, it was probably broken since. I might try switching to macOS Perl as the Perl dependencies are pre-installed (may not be possible depending on bioperl). Relying on cpanminus may have made the build harder to reproduce.

man-db failed in previous PR too. Will need to determine if it needs a revision bump. Test failure output isn't very useful.

[cho-m]

pidgin is actually due to intltool failure (XML::Parser is a compiled library)

$ PERL5LIB=/home/linuxbrew/.linuxbrew/opt/intltool/libexec/lib/perl5 perl -e "require XML::Parser"
Expat.c: loadable library and perl binaries are mismatched (got first handshake key 0xeb00080, needed 0xeb80080)

$ file /home/linuxbrew/.linuxbrew/opt/intltool/libexec/lib/perl5/x86_64-linux-thread-multi/auto/XML/Parser/Expat/Expat.so
/home/linuxbrew/.linuxbrew/opt/intltool/libexec/lib/perl5/x86_64-linux-thread-multi/auto/XML/Parser/Expat/Expat.so: ELF 64-bit LSB shared object, x86-64, version 1 (SYSV), dynamically linked, BuildID[sha1]=ad3c15e5bc2cbc090fb4422af4e1e72ab735677e, not stripped

[cho-m]

Alternatively could have

• manually moved files (man3.install ...) or
• modified the Makefile

  inreplace "Makefile", "$(PERL) Makefile.PL PREFIX=$(prefix)", "\\0 INSTALLSITEMAN3DIR=#{man3}"

The current approach is officially documented method for adding options during Perl build https://github.com/apache/subversion/blob/1.14.2/subversion/bindings/swig/INSTALL#L217-L236

PREFIX is from the Makefile https://github.com/apache/subversion/blob/1.14.2/Makefile.in#L883-L884 while INSTALLSITEMAN3DIR is to fix the manpage installation.

[danielnachun]

The documented approach looks pretty clean so it seems fine to me.

[cho-m]

Hopefully, only remaining failure will be Linux grokj2k. I don't think we can fix in this PR so may need to follow up afterward.

[danielnachun]

Agreed, we should follow up with this in a PR to upgrade grokj2k.

[cho-m]

Working on it in #112269. Current problem is macOS Perl headers conflict with C++ headers. It builds for me if I disable exiftool feature. Need to look into it a bit more:

/Library/Developer/CommandLineTools/SDKs/MacOSX12.sdk/usr/include/c++/v1/locale:3569:35: error: too few arguments provided to function-like macro invocation
        return do_open(__nm, __loc);
                                  ^
/Library/Developer/CommandLineTools/SDKs/MacOSX12.sdk/System/Library/Perl/5.30/darwin-thread-multi-2level/CORE/perl.h:6924:9: note: macro 'do_open' defined here
#define do_open(g, n, l, a, rm, rp, sf) \
        ^

[cho-m]

I think this is ready to merge once CI finishes. macOS is all passing while Linux currently only failed on grokj2k, which should be fixed in #112269.

[danielnachun]

Everything is passing on macOS and as hoped the only Linux failure is grokj2k. Let get this merged.

[BrewTestBot]

@danielnachun has triggered a merge.

[BrewTestBot]

⚠️ @danielnachun bottle publish failed.

bottle publish failed

[BrewTestBot]

@danielnachun has triggered a merge.