Properly fix CI

[cooljeanius]

CI systems now have incompatible BFD headers installed on them, which get picked up during the build, and cause it to fail. We'll need to either find a way to ignore them, add compatibility with them, or just not install them in the first place?

[cooljeanius]

ok well it's "fixed" (i.e. hacked around) on Travis now; won't know about GitHub Actions until those actually decide to run again, though...

[cooljeanius]

"fixed" now on GitHub Actions, too, but only by doing the same hacking-around-the-problem that I did on Travis, so I'm leaving this open until I come up with a proper fix...

[cooljeanius]

Current failures being ignored and/or hacked around are:

• Linux: bfd can't be built because we need to find out how ksigmask_t is properly defined (fixed by 3592bfa)
• top-level Makefile: need xcrun to be available under Darling (which itself needs to be setuppable on CI)
• libcheckpoint: gnutar is no longer a developer tool recognized by xcrun (fixed by da910d1)
• libcheckpoint: missing CG (CoreGraphics?) prefixed functions; probably were deprecated and removed at some point...
• MacsBug: build should be attempted after bfd's headers are built so that bfd.h is available
• ant build: libiberty needs to be able to find config-ml.in properly when built outside of its srcdir
• gradle build: just generally broken; doesn't do any of the things it's supposed to... (CodeQL mentions a failure to specify a version number, scan builds require accepting user agreement, etc.)
• normal Mac build: binutils can't be built due to headers failing to be re-generated / re-created properly duplicate symbol error for yyparse:

duplicate symbol '_yyparse' in:
    arparse.o
    ../intl/.libs/libintl.a(plural.o)
ld: 1 duplicate symbol for architecture x86_64
clang: fatal error: linker command failed with exit code 1 (use -v to see invocation)
make[3]: *** [ar] Error 1
make[2]: *** [all-recursive] Error 1
make[1]: *** [all] Error 2
make: *** [all-binutils] Error 2

• a bunch of testsuites have to be skipped
• CodeQL runs out of memory for 'c-cpp'. Apparently it needs a larger runner, which it looks like I'd have to pay for... (fixed by e263c65)
• Homebrew dependencies have recently started to conflict with stuff that's preinstalled on the default macOS runners; attempting to fix this in PR Fix Homebrew dependencies for CI #47
• "Full build to get anything that might have been missed" step: mv: rename libltdl/loaders/.deps/dlopen.Tpo to libltdl/loaders/.deps/dlopen.Plo: No such file or directory
• "Full build to get anything that might have been missed" step: /Applications/Xcode_15.0.1.app/Contents/Developer/Toolchains/XcodeDefault.xctoolchain/usr/bin/gm4: unrecognized option '--gnu'

[cooljeanius]

(note on the checklist: fixing the ksigmask_t one kinda caused the "CodeQL OOM" one...)

[cooljeanius]

I think I've made some progress on the last 2 items on the checklist, and have checked them off for now...

[cooljeanius]

Update on the last one, about autoconf trying to pass the --gnu flag to gm4 even when it doesn't recognize it: I reported that to the upstream bug-autoconf mailing list here: https://lists.gnu.org/archive/html/bug-autoconf/2024-07/msg00003.html