llvm-packages: get rid of extra build depedencies for manpages #26818
Conversation
|
@LnL7, thanks for your PR! By analyzing the history of the files in this pull request, we identified @dtzWill, @copumpkin and @acowley to be potential reviewers. |
LnL7
added
the
1.severity: mass-rebuild
| @@ -1,4 +1,4 @@ | |||
| { newScope, stdenv, cmake, libxml2, python2, isl, fetchurl, overrideCC, wrapCC, darwin, ccWrapperFun }: | |||
| { lowPrio, newScope, stdenv, cmake, libxml2, python2, isl, fetchurl, overrideCC, wrapCC, darwin, ccWrapperFun }: | |||
There was a problem hiding this comment.
I could be missing something but there's no reference to lowPrio
There was a problem hiding this comment.
Indeed, forgot to apply it to the *-manpages attributes
|
Looks good, although it's a pity we have to do a whole new 1h+ build to get some documentation. Someone who understands the LLVM build internals a bit better can probably optimize that a bit though. |
|
Should this go against staging? Or shall we just take the mass-darwin-rebuild? |
|
Thanks @LnL7! |
|
@LnL7: this makes all darwin stuff broken, so Hydra hasn't been building anything for it on staging. Can you/someone fix this fast? (I don't see directly what packages cause it.) Otherwise I suggest to revert for now and leave it for another staging iteration. |
|
Are you sure it was caused by these changes? I merged this in a week ago and tested it with my local hydra setup. |
|
Oh, I'm sorry, I messed this up. The real culprit is 4ac1901. (Fortunately, I can evaluate easily without having a Darwin machine.) |
|
This breaks the NixOS channel update: The reason is that this line: makes Hydra think that the A possible fix is to have |
|
Hmm, maybe this approach to override We certainly want to unblock the channel fast – any immediate ideas? If not, I'd probably mark this |
|
Easiest fix is to remove |
|
Oh, right. I thought of |
It was blocking channel(s). See: #26818 (comment)
|
Pushed as 1e1472e. |
|
The bigger concern for me is if |
|
The main UX blocker ATM for multiple outputs (e.g. man) is #24717, I think. |
|
@edolstra: there's apparently some other blocker I can't see, as that very commit has successfully passed the tested job and there's still no channel bump. |
|
Maybe it's related to Hydra's evaluator complaining about being out of space? |
|
Both nixos.org and Hydra has a disk full. Both seem fine now. However, the Looks like the channel would go back in time. Maybe some builds got restarted? |
|
Yes, I restarted the failing builds as an attempt to get this going forward. I don't understand why channel would go back in time, since it's 13 days old? |
|
@edolstra: nixos-unstble branch in nixpkgs-channels is at a rather new commit 1e1472e that doesn't correspond to what's in http://nixos.org/channels/nixos-unstable. I suspect all this is due to non-atomicity of the mirroring script (in combination with some error). |
|
From earlier in the journal: So it updated the |
rrbutani
added
the
6.topic: llvm/clang
Motivation for this change
Fixes #26756
Things done
(nix.useSandbox on NixOS,
or option
build-use-sandboxinnix.confon non-NixOS)
nix-shell -p nox --run "nox-review wip"./result/bin/)