haskellPackages.selda-sqlite: unbreak#405820
haskellPackages.selda-sqlite: unbreak#405820rnhmjoj wants to merge 1 commit intoNixOS:haskell-updatesfrom
Conversation
github-actions
bot
added
the
6.topic: haskell
github-actions
bot
added
10.rebuild-darwin: 1-10
There was a problem hiding this comment.
I have a weak preference for substituteInPlace --replace-fail since that would let us know when to remove this.
FTR, we intentionally leave conditionals alone NixOS/jailbreak-cabal@99eac40.
There was a problem hiding this comment.
FTR, we intentionally leave conditionals alone NixOS/jailbreak-cabal@99eac40.
I'm not sure I'm buying the argument made in that commit. The constraints under the flag are required to pick the correct version for a solver - and then to add another peer dependency at the same time for this case.
But none of that applies to us in nixpkgs. The versions we provide are set. The flag configuration is as well. When you hit a constraint and want to jailbreak it, it doesn't matter for us whether it's behind a flag or not.
What am I missing?
Edit: Maybe I am missing some kind of reverse mechanism that sets the flags based on the provided dependencies? I guess I just don't know about this feature.
There was a problem hiding this comment.
A flag conditional may, besides restricting the version ranges, set any number of things including e.g. compilation flags. If the flag is marked as automatic, Cabal will attempt to automatically set it based on the preinstalled (by us) versions of packages. If we remove those constraints, it just has to go with the default value of the flag, which may mean incorrect settings and the build failing incorrectly, e.g. because a CPP var isn't set.
There was a problem hiding this comment.
I have a weak preference for substituteInPlace --replace-fail since that would let us know when to remove this.
It's not possible with substituteInPlace: this code applies to 3 packages and the whitespace is different in each .cabal file, so you need a regex. However we can make sed fail too.
There was a problem hiding this comment.
So, I guess 3x replace-fail would be easiest to maintain for us (?)
rnhmjoj
force-pushed
the
pr-fix-selda-sqlite
branch
from
e000bb5 to
cc96b05
Compare
There was a problem hiding this comment.
That's just gonna have the build fail without any message which will cost whoever is going to debug this some time…
rnhmjoj
force-pushed
the
pr-fix-selda-sqlite
branch
from
cc96b05 to
6d7682b
Compare
|
Picked as 802c06b. If we were only able to stop reviving these dead projects… |
Someone is maintaing a fork, perhaps we should switch upstream? |
|
I would rather suggest to them doing a Hackage takeover request if they're serious. Their talk of “company specific fixes” is not too confidence inspiring. |
3 participants
It seems the cabal file of this package is not handled properly by jailbreak-cabal.
It has to be modified manually to remove the constraints.
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.