-
Notifications
You must be signed in to change notification settings - Fork 163
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Workaround for ghcjs hardcoding to linux #511
base: develop
Are you sure you want to change the base?
Conversation
else nixpkgs.haskell.lib.overrideCabal package (drv: { | ||
postPatch = (drv.postPatch or "") + nixpkgs.lib.optionalString (system == "x86_64-darwin") '' | ||
substituteInPlace ${cabalFile}.cabal --replace 'if os(linux)' 'if os(linux) && !impl(ghcjs)' | ||
substituteInPlace ${cabalFile}.cabal --replace 'if os(osx)' 'if os(linux) && impl(ghcjs)' |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Is this right? if os(osx)
gets translated to if os(linux) ...
?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
If it is right...an explanation is worthy.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Talked to @alexfmpe. So basically this is just a very confusing way of writing it. Because of the hardcoding in GHCJS, os(linux) == true
and true && x == x
so why are we confounding this substitution?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
We don't want the os(linux)
patches to get applied, so we change the os(linux)
conditional to only apply when GHCJS isn't being used. Since GHCJS always thinks it's running on Linux, we change the os(osx)
patches to os(linux)
patches.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
because we don't want to apply the actual os(linux)
patches (since we're really on macOS)
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
GHCJS thinks we're on Linux, but that doesn't mean we should apply the Linux patches, instead we need to trick GHCJS into applying the macOS patches
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
so we say "these linux patches are not for you, silly GHCJS, but these other (actually macOS) linux patches are!"
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Note the replacement only happens if system == "x86_64-darwin"
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
That makes sense. Thanks.
As @3noch said, this needs a more informative comment/explanation. Rather than substitute-in-place, please create a patch file and apply it. Please also document where the upstream PR we're waiting on can be found. |
ghcjs/ghcjs#702 mentions the issues. |
Due to ghcjs/ghcjs#674
Upstreaming @vaibhavsagar's fix since we both ran into this on different projects.
Test case: obsidiansystems/obelisk@develop...alexfmpe:ae-error-foundation