justStaticExecutables: Forbid references to GHC #304352
Conversation
9999years
requested review from
sternenseemann,
maralorn and
ncfavier
as code owners
|
I like this. A bit sad that this makes justStaticExecutables better than enableSeparateBinOutput. I forgot whether there is a per output option regarding requisites. We will need to make sure that this doesn’t break any of the Haskell binaries that we ship in nixpkgs so this definitely needs to merge into haskell-updates. |
|
@maralorn Agreed, how do I do that? Just change the merge target? |
There was a problem hiding this comment.
disallowedRequisites needs to become an argument to haskellPackages.mkDerivation and transparently passed on to stdenv.mkDerivation in there. Otherwise disallowedRequisites will only be set depending on override order, i.e. any overrideCabal-based function applied after justStaticExecutables would revert the change again, as it uses a different fix point from overrideAttrs.
|
@sternenseemann I'm not sure how to make that work; I can add Maybe it could be made to work with some functor magic, but I wouldn't know how to wire it up robustly. |
|
@9999years |
|
Ah, I get what you mean. In principle you can literally add an |
|
Or |
9999years
force-pushed
the
just-static-executables-disallow-ghc
branch
from
49ce66f
to
dc0af03
Compare
|
OK, I took a shot at it, let me know what you think. I also checked that the docs render correctly: 
|
ofborg
bot
added
10.rebuild-darwin: 501+
10.rebuild-darwin: 5001+
10.rebuild-linux: 5001+
and removed
10.rebuild-darwin: 101-500
10.rebuild-linux: 501-1000
labels
|
@sternenseemann Still looking for a review on this! |
9999years
force-pushed
the
just-static-executables-disallow-ghc
branch
from
dc0af03
to
e1bc2ba
Compare
|
PTAL |
This change ensures that packages won't be rebuild compared to before the introduction of disallowedRequisites/disallowGhcReference unless they use one of those arguments. It may be nice to revert this in the future (via staging) for greater simplicity, but will help with initial regression testing.
sternenseemann
force-pushed
the
just-static-executables-disallow-ghc
branch
from
d83abc3
to
9e9a2f2
Compare
sternenseemann
removed
the
9.needs: port to stable
|
I've realized that we can't backport this anymore since this is technically a breaking change. I've moved the changelog entry to the 24.11 release notes as a result. We could backport a version of this which adds the new arguments to I've also noticed that we haven't documented |
ofborg
bot
added
10.rebuild-darwin: 101-500
10.rebuild-linux: 501-1000
and removed
10.rebuild-darwin: 501+
10.rebuild-darwin: 5001+
10.rebuild-linux: 5001+
labels
|
@9999years Evals newer than 1806586 of haskell-updates contain your change. There are already a few regressions I've seen on aarch64-darwin. This is not really surprising as GHC can't properly eliminate dead code on that platform. It may be prudent to ignore |
Which attributes are impacted? |
|
@9999years They should accumulate here. Note that a lot of builds are still pending so the list will get longer as time goes on. |
Fixes `cabal-install` to remove references to GHC, mostly through other libraries that are included in the binary. See: NixOS#304352
Remove references to `hpack` and `distribution-nixpkgs` paths so that `cabal2nix` can build on macOS. See: NixOS#304352
Remove references to `hpack` and `distribution-nixpkgs` paths so that `cabal2nix` can build on macOS. See: #304352
Fixes `cabal-install` to remove references to GHC, mostly through other libraries that are included in the binary. See: #304352 Co-authored-by: sternenseemann <[email protected]>
All these references would (indirectly) incur a GHC requisite which is prevented by NixOS#304352 via justStaticExecutables. Consequently, we can stop setting disallowedReferences. As it turns out the references we were removing aren't currently created, so our life gets even easier.
All these references would (indirectly) incur a GHC requisite which is prevented by #304352 via justStaticExecutables. Consequently, we can stop setting disallowedReferences. As it turns out the references we were removing aren't currently created, so our life gets even easier.
This commit disables justStaticExecutables for packages on aarch64-darwin that incur a (usually incorrect) reference on GHC. justStaticExecutables now fails when a GHC reference remains in the output (#304352). Due to this reference justStaticExecutables (before these changes) would only marginally reduce closure size. In the future, we'll be able to re-introduce justStaticExecutables after (manually) removing the incorrect references stemming from the use of Paths_* modules. Tracking Issue: #318013
Description of changes
This makes
justStaticExecutableserror if the produced store path contains references to GHC. This is almost always erroneous and due to the generatedPaths_*module being imported. This helps preventjustStaticExecutablesfrom producing binaries with closure sizes in the gigabytes.See: #164630
Things done
nix.conf? (See Nix manual)sandbox = relaxedsandbox = truenix-shell -p nixpkgs-review --run "nixpkgs-review rev HEAD". Note: all changes have to be committed, also see nixpkgs-review usage./result/bin/)Add a 👍 reaction to pull requests you find important.