Make hash-breaking cleanups avoided in #26007 #26881
Conversation
|
What kind of prefix is |
Ericson2314
force-pushed
the
cross-hashbreak
branch
2 times, most recently
from
286b612
to
75bfc10
Compare
|
@FRidh Added some comments for that. |
| @@ -17,7 +16,7 @@ stdenv.mkDerivation rec { | |||
| sha256 = "c3e5e9fdd5004dcb542feda5ee4f0ff0744628baf8ed2dd5d66f8ca1197cb1a1"; | |||
| }; | |||
|
|
|||
| postPatch = stdenv.lib.optionalString stdenv.isDarwin '' | |||
| postPatch = stdenv.lib.optionalString hostPlatform.isDarwin '' | |||
| substituteInPlace configure \ | |||
| --replace '/usr/bin/libtool' 'ar' \ | |||
There was a problem hiding this comment.
An impurity like /usr/bin/ seems worth patching out regardless of the platform, right?
There was a problem hiding this comment.
Then again, you aren't really changing this logic here, so I guess it's not worth fixing here.
There was a problem hiding this comment.
@copumpkin you're looking at the person that just ran `s/stdenv.is/hostPlatform.is/g' :D.
In seriousness, looks like libtool is being replaced with ar, which is kind of sketchy? I'd guess:
libtoolshould continue to be used where possible, though certainly not impurelyarshould be replaced with$AR- Is this impurity still there upstream? The Linux build doesn't fail with sandboxing at least, and I doubt the e.g. homebrew people want that hard-coded.
Ericson2314
added
6.topic: cross-compilation
|
Hydra seems to not want to evaluate the job? |
Ericson2314
force-pushed
the
cross-hashbreak
branch
2 times, most recently
from
cfef6d4
to
c33e29f
Compare
|
At least the cross tests still work, which ends up needing the guts of the mass rebuild just fine. Native coreutils has this test https://github.com/coreutils/coreutils/blob/master/tests/install/install-C.sh fails for some reason, however:
|
Ericson2314
force-pushed
the
cross-hashbreak
branch
2 times, most recently
from
4053af1
to
6875845
Compare
|
coreutils built after rebuild now. |
Ericson2314
force-pushed
the
cross-hashbreak
branch
3 times, most recently
from
46bc85d
to
5b766ba
Compare
PR NixOS#26007 used these to avoid causing a mass rebuild. Now that we know things work, we do that to clean up.
Ericson2314
force-pushed
the
cross-hashbreak
branch
from
5b766ba
to
95c8277
Compare
|
hydra.nixos.org kept on timing out, but I built the Travis commands locally on both Linux and Darwin, which should build all changed platforms. so merging to staging. |
Motivation for this change
PR #26007 used these to avoid causing a mass rebuild. Now that we know things work, I'm doing a cleanup of all the individual derivations. Beyond tidiness for tidiness's sake, I think this is very important because people write packages by copying existing packages, so it's important to demonstrate the intended idem.
Some of the core infrastructure also needs a cleanup, but I'm leaving that in #26805 as it will take more effort to get correct
Things done
mass rebuild tests: http://hydra.nixos.org/jobset/nixpkgs/ericson2314-cross-staging
cross tests: http://hydra.nixos.org/jobset/nixpkgs/ericson2314-cross-scratch
CC @Dridus @copumpkin