revert haskell.nix update #503
Conversation
angerman
force-pushed
the
angerman/unbreak-ci
branch
from
066dd24
to
7aa4bd2
Compare
|
@lehins this is as green as we can get it right now, I believe. There are only some 9.8.2 windows cross build failures, everything else is green. |
There was a problem hiding this comment.
Looking at the changes in the PR makes me believe that something is broken on haskellNix and we are waiting for it to be fixed. Only after it is fixed then we restore using newer ghc and cabal. Is that correct?
If my understanding is correct, could you please link to a github issue that tracks this, so we know when we can get rid of the git pin for haskellNix.
Also, could I ask you to cleanup the git history in this PR a little bit. If you won't get it to it, not the end of the world, I'll squash all them together when merging.
| @@ -48,7 +56,7 @@ | |||
| inherit (nixpkgs) lib; | |||
|
|
|||
| # see flake `variants` below for alternative compilers | |||
| defaultCompiler = "ghc966"; | |||
| defaultCompiler = "ghc965"; | |||
| @@ -84,7 +92,7 @@ | |||
| # tools we want in our shell, from hackage | |||
| tools = | |||
| { | |||
| cabal = "3.12.1.0"; | |||
| cabal = "3.10.3.0"; | |||
There was a problem hiding this comment.
Same here, how come we need to downgrade cabal?
| -- Windows cross with bitvec/simd is broken. | ||
| package bitvec | ||
| -- Workaround for windows cross-compilation | ||
| flags: -simd |
There was a problem hiding this comment.
Do you know if this is a valid syntax for cabal.project? Never tried it before, but if it works it would be much better to turn it off only on platforms that are problematic. The concern I have is that if we do use simd on linux and mac for the actual cardano-node then all of the tests in this repo are not tested with the same flags
| flags: -simd | |
| if os(windows) | |
| flags: -simd |
This hopefully unbreaks the CI issues we see.