9.14 support #4780
9.14 support #4780
Conversation
wz1000
commented
Jan 1, 2026
wz1000
commented
- wip: Build ghcide with 9.14
- plugins build with 9.14
wz1000
requested review from
July541,
drsooch,
fendor,
joyfulmantis,
michaelpj and
santiweight
as code owners
wz1000
force-pushed
the
wip/9.14
branch
2 times, most recently
from
43a20ad to
0deb1d2
Compare
| source-repository-package | ||
| type: git | ||
| location: https://github.com/snowleopard/alga | ||
| tag: d4e43fb42db05413459fb2df493361d5a666588a |
There was a problem hiding this comment.
Do we need this? Or is this just an allow-newer?
There was a problem hiding this comment.
I advice against using source-repository-package. I think it is a workaround when the package is not yet available on another repository such as Hackage. Can we update alga on Hackage instead?
There was a problem hiding this comment.
Unfortunately, alga requires changes and can't be revised on hackage. While this is rather unfortunate, we will have to release with this source-repository-package stanza, as we are otherwise blocked for an unknown amount of time.
There was a problem hiding this comment.
@fendor It's perfectly fine to vendor the patched alga in the HLS tree for the time being and remove it after the fix has been released. This allows people to install HLS from its Hackage sdist without trouble. :)
| description: "Cabal version" | ||
| required: false | ||
| default: "3.14.2.0" | ||
| default: "3.16.0.0" |
There was a problem hiding this comment.
This should be bumped to 3.16.1.0
|
FYI Building HLS 2.12.0.0 using GHC 9.12.3 fails, too:
Will this PR also fix this? |
|
@benz0li I don't think so, the failed CI jobs point to |
|
According to https://discourse.haskell.org/t/critical-code-generation-bug-with-ghc-9-12-3/13505 we should avoid using GHC 9.12.3 altogether. (Maybe irrelevant). |
True. Then, I wonder why it picks |
|
@benz0li Ah, perhaps the index-state needs to be updated? That might be fixed by this PR. |
Cross reference: |
