v0.9.11 breaks texlive on darwin

[nim65s]

Hi again,

Not sure if this is the same issue as #498 or not, but on nixpkgs and darwin, when we use #497 and ignore #498, libfaketime returns err 137, at least when use it to build texlive:

texlive> fmtutil: running `faketime -f '@1980-01-01 00:00:00 x0.001' luatex -ini   -jobname=dvilualatex -progname=dvilualatex dvilualatex.ini' ...
texlive> fmtutil [ERROR]: running `faketime -f '@1980-01-01 00:00:00 x0.001' luahbtex -ini   -jobname=luahbtex -progname=luahbtex luatex.ini </dev/null' return status: 137
texlive> fmtutil [ERROR]: returning error due to option --strict

cc. @usertam, as I understood that you are working on this already

[usertam]

Hello! For context I was the person who did the 0.9.10 update in nixpkgs: NixOS/nixpkgs#206860, but at that time didn't add myself to maintainers. A few days ago I was trying to do the same for 0.9.11, but I realized that most things broke and there are no good easy fixes, so I filed this: #497, and did not bother to touch anything in nixpkgs.

The PR NixOS/nixpkgs#412913 should never be merged in the first place since the release is broken and nothing is tested on platforms other than x86_64-linux, but I wasn't a maintainer so I wasn't aware.

As of now, the aarch64-darwin build will crap itself and throw Trace/BPT trap: 5 when being loaded by dyld. It's completely broken.

And as I said already in #497, that PR alone will NOT fix building on darwin. This commit 2503b0f must be reverted or changed in some way that does not break darwin, or BUG IN CLIENT OF LIBPLATFORM: Trying to recursively lock an os_once_t will continue to happen.

I will continue to work on this when I am free during the weekend, but anyone is welcome to file a PR to fix the above issue. And thanks for pinging me about the issue. :D

[usertam]

Hello, for interested people can you test #500? I am hoping it will fix this indirectly 🙏

Edit: I also created a PR at NixOS/nixpkgs#414782 with the fixes included.