devShell is not being built/tested by CI #383

New issue

Closed

opened 2024-06-09 13:42:32 +00:00 by delroth · 3 comments

delroth commented

2024-06-09 13:42:32 +00:00

Member

28a079f841 recently regressed nix develop in a very obvious way and this had to be caught manually after the fact. The CI should at the very least try to eval and build the devShell target.

28a079f841834ac11105ce5762115904d1e40ce1 recently regressed `nix develop` in a very obvious way and this had to be caught manually after the fact. The CI should at the very least try to eval and build the devShell target.

delroth added the

devx

label 2024-06-09 13:48:50 +00:00

jade commented

2024-06-09 15:31:19 +00:00

Owner

ci task should perhaps be nix develop the target inside a derivation then run a trivial command or like maybe git stuff. the only problem is that nix develop actually creates a new derivation because it's very bad, and so we might have to replicate the "adding -env to the name gratuitously" to make that a cache hit to not require store write access.

qyriad commented

2024-06-09 20:39:50 +00:00

Owner

At the very least we can test simply evaluating and building the dev shell derivation in hydraJobs. That doesn't test that things in the shell work, but it would have caught this particular regression

At the very least we can test simply evaluating and building the dev shell derivation in hydraJobs. That doesn't test that things in the shell *work*, but it would have caught this particular regression

👍 1

delroth referenced this issue

2024-06-10 10:12:16 +00:00

make ci fail if any hydra job does not eval #237

lix-bot commented

2024-06-11 20:17:54 +00:00

Member

This issue was mentioned on Gerrit on the following CLs:

commit message in cl/1438 ("Check devShells in CI")

This issue was mentioned on Gerrit on the following CLs: * commit message in [cl/1438](https://gerrit.lix.systems/c/lix/+/1438) ("Check devShells in CI")