llvm*Packages: fix output selection (lib.get*) #122554
Conversation
|
It's fallout from PR #111487. Well, maybe the old generic logic around |
|
Perhaps my mind is flakey. It might also have purpose in |
|
@GrahamcOfBorg build ruby |
|
Result of 465 packages marked as broken and skipped:
8494 packages skipped due to time constraints:
2 packages built successfully:
Result of 409 packages marked as broken and skipped:
2055 packages skipped due to time constraints:
31 packages built successfully:
|
|
Hmm, no luck with verifying ruby on OfBorg – it timed out. I expect the change also caused rebuild of some of its deeper dependencies. |
|
@vcunat I think It might also help to do |
|
You could, but so far our general convention is to choose the output with binaries as the default one; IIRC only glibc was an exception 😄 |
|
@vcunat well, I defer to you on that as original author of multiple outputs in Nixpkgs :). |
|
@vcunat anyways I think more |
|
I clearly managed to forget a lot of the context since then, so I'm probably not really special in this respect (anymore). I suppose we'll try it then; perhaps I'll wait a bit in case someone else wants to have a look, too. |
|
... or not 😄 I'm feeling optimistic and Hydra would otherwise be quite idle now. It's just staging-next, so in the worst case we can change it again. |
The effect of `.out // { outputSpecified = false; }` in these cases
is to select the default output explicitly, but then make the
selection implicit until `overrideAttrs` is called. Previously
`overrideAttrs` would not preserve output selection, masking the
apparently unnecessary behavior of this workaround.
For `libllvm-polly`, this logic does not apply, as it does not
select the default output.
The `outputSpecified` workaround was introduced in
NixOS#122554
and was perhaps rushed because of a release deadline, and expected
delays from mass rebuilds.
The change in `overrideAttrs` behavior was added in
5b2f597 / NixOS#211685
and the problem was discovered in NixOS#218537,
which may contain further information.
The effect of `.out // { outputSpecified = false; }` in these cases
is to select the default output explicitly, but then make the
selection implicit until `overrideAttrs` is called. Previously
`overrideAttrs` would not preserve output selection, masking the
apparently unnecessary behavior of this workaround.
For `libllvm-polly`, this logic does not apply, as it does not
select the default output.
The `outputSpecified` workaround was introduced in
#122554
and was perhaps rushed because of a release deadline, and expected
delays from mass rebuilds.
The change in `overrideAttrs` behavior was added in
5b2f597 / #211685
and the problem was discovered in #218537,
which may contain further information.
rrbutani
added
the
6.topic: llvm/clang
Motivation for this change
#120058 (comment)
Things done
sandboxinnix.confon non-NixOS linux)nix-shell -p nixpkgs-review --run "nixpkgs-review wip"./result/bin/)nix path-info -Sbefore and after)