-
Notifications
You must be signed in to change notification settings - Fork 237
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
withHoogle
does not respect exactDeps
for hoogle itself
#2277
Comments
There is upstream issue in hackage that breaks snoyberg/http-client#547 To workaround this issue you can override the
This will pin the |
thanks for the workaround tip with cabal picking whatever it likes still though this makes general reproducability more troublesome with cached results differing from uncached. Is there any scope for this being able to be pinned like exactDeps in the future? |
I don't think haskell.nix should be responsible for pinning tools like Haskell.nix has a source pin for Haskell.nix projects can pin
That should globally limit the Something else I should mention.
Currently there is no way to choose
|
Thanks for the info |
* Pin index-state used to build hoogle in tests See #2277 * Also fix shell-for test * Bump index-state to 2024-10-26
This is required due to this mess: snoyberg/http-client#547 A different fix is required for Hoogle in the `nix develop` shell. See: input-output-hk/haskell.nix#2277 Signed-off-by: Drew Hess <[email protected]>
It looks like the underlying issue here is that If we add add
Adding |
* Update nixpkgs and ghc HEAD * Fix warning * Update internalHackageIndexState * Update patch bounds * Update rcodesign for recent MacOS * Replace ghc911 with ghc912X and ghc913 * ifdLevel 1 * Use lazy-inputs instead of fetch-source overlay * Skip broken mingwW64 GHC * Skip broken mingwW64 GHC * Add GHC 9.8.3 * Update patch bounds * More patch fixes * Turn off pool allocator in favour of m32_alloc solution in ghc 9.8.3 * Remove another redundant patch * Turn off pool allocator in favour of m32_alloc solution in ghc 9.8.3 * Remove another redundant patch * Fix ghc 9.10 * Update materialized files * Update materialized files * Build ghc983 cross compilers with ghc983 * Patch upper bounds * ifdLevel 2 * Disable broken tests * ifdLevel 3 * Fix cabal-latests * Update head.hackage * Disable HLS for ghc 9.8.3 for now * Update nixpkgs-unstable and fix issue with no libcxxabi * Bump ghcjs-overlay and head.hackage * Enable external-static-plugin test for GHC >=9.6 (Fixes #2265) * Bump head.hackage * Fix external-static-plugin test * Fix external-static-plugin test * Fix eval issue with latest nixpkgs * Skip th-dlls test for ghc 9.12 and head brances * Fix HLS test * Fix hls test * Bump head.hackage * Bump head.hackage * Bump GHC git pins * ifdLevel 2 * ifdLevel 3 * Add hackage-quirk for #2277 * Bump head.hackage * Add hackage-quirk for #2277 * Fix for ghcjs on macOS * Disable broken tests * Disable broken test * Fix update scripts to use nixVersions.stable
This has been fixed in hackage. #2279 removes the workaround from haskell.nix. |
Describe the bug
we have a nix shell that looks very similar to the example
https://input-output-hk.github.io/haskell.nix/tutorials/development.html#how-to-get-a-development-shell
We use stackage for snapshots and currently target
nightly-2024-08-08
which importantly contains a working combination of data-default and http-client-tlshttps://www.stackage.org/nightly-2024-08-08/package/data-default-0.7.1.1
https://www.stackage.org/nightly-2024-08-08/package/http-client-tls-0.3.6.3
However if we enable
withHoogle: true
in the last few days we are unable to build this as it fails to compile with the combination ofhttp-client-tls-0.3.6.4
anddata-default-0.8.0
neither of which are in our snapshot.It seems that cabal itself is deciding which dependency versions to use which I was expecting
exactDeps
to stopExpected behavior
hoogle server should respect exactDeps or at least the snapshot
** Extra context **
This is the error you get currently when cabal picks the latest versions
The text was updated successfully, but these errors were encountered: