As I put on the issue, I have tested that this fixes the problem I had with biber. I haven't installed full texlive as it would require another llvm compile but can't see why it would be any different.

Maybe that's unrelated to this change https://hydra.nixos.org/build/92065248

Oh I'm very glad this sort of thing was the cause--I've been experiencing all manner of perl shebang madness and it was so close but still requiring attention in a few places (such as biber).

Will pull in and test a bit, don't delay on my account but I'll hopefully report back with some positive news once in a $rebuild_completion_period or so. ⌚

Maybe that's unrelated to this change https://hydra.nixos.org/build/92065248

Yes, it was broken in April but already fixed in #60861

This pr clearly doesn't have that commit 0a64e79.
Thinking maybe I should check if this is still an issue on staging then.

ping @volth . Any insights?

So from what you said this is an atypical issue and any anticipated regressions would be limited as it's limited already. 👍 Making it less of an unattractive "fix".

Hmm and not having it causing issues for others right away dhbw-stginf16a/nosql-book#23 🤣

I am hoping we can also revert 5a5a597 with this

Can this go in?

Can this go in?

Perhaps just
75420af

to avoid mass-rebuild and possible affecting 1000+ of perlPackages
The rest of derivations (ddclient, po4a) are already fixed, so biber seems to be the only one which need hostPlatform's perl

Anything that installs a perl script that needs to be executed will need the host platform's perl. That should be a lot of scripts that it isn't worth patching these manually. We should revert those fixes.