Commit
This commit does not belong to any branch on this repository, and may belong to a fork outside of the repository.
Revert "Try not caching journey test dependencies"
This is part of investigating the problem of not being able to connect to the git daemon. This problem is described in: GitoxideLabs#1634 (review) The preceding commit, bf2abdb, which this reverts, omitted caching from the `test-journey` CI job, and that somehow avoided the problem, causing all tests to pass. With that change, the failure did not recur even after rerunning the job more than 15 times: https://github.com/EliahKagan/gitoxide/actions/runs/12363538568 That is strange, because no cause related to Rust dependencies is apparent: The Git daemon is not itself provided by or expected to be affected by them, and any stale dependencies from the cache should automatically not be used. However, perhaps the extra *delays*, incurred by not having cached some artifacts and having to rebuild them, allowed the failure-triggering condition to resolve on its own. If so, that would explain this observation. It would also be consistent with how the problem arose with the changes in GitoxideLabs#1634, if a previous delay existed that happened to prevent the bug from being observed and that GitoxideLabs#1634 somehow counteracted. This could be either by a performance improvement, or by shifting the delays to other times during the job (by changing which artifacts were needed for which tests). The `test-journey` job usually takes between 6:00 an 6:30 without caching build artifacts. This is arguably not too bad, so it might be feasible to use that change as a long-term workaround. But if possible the underlying problem should be identified and fixed. So this commit brings back caching to check if it reliably restores the bug. If so, then experiments and other fixes or workarounds can be attempted on top of that.
- Loading branch information