rworkflows
rworkflows copied to clipboard
Force new install of local repo
I noticed with autoCV that the local repo wasn't getting reinstalled when the version was detected to be the same as before (0.99.0).
https://github.com/bschilder/autoCV/actions/runs/6484810363/job/17609460418#step:4:599
Skipping install of 'autoCV' from a local remote, the SHA1 (0.99.0) has not changed since last install.
Use `force = TRUE` to force installation
Bumping the version to 0.99.1 fixed this, but I think the default behaviour should be to always reinstall, because developers aren't going to necessarily bump the version every time they make a push.
Done in dev:
https://github.com/neurogenomics/rworkflows/blob/059489ca98b668255452256ecbbbc06a7be59cef/action.yml#L306
Hey Brian, any reason why this was reverted to force = FALSE? CAnnot figure out why, but it seems to cause trouble again...
There were a couple of reasons, and I can see arguments for either as a default. but i think the real answer is that users should have control over this param, which ive implemented on rworkflows dev but havent released yet.
some pros of force=FALSE:
- speeds up GHA runs and reduces unnecessary reinstalls.
- one user noted they had a suggest that couldnt be installed, and they wanted to pass GHA anyways: https://github.com/neurogenomics/rworkflows/issues/117
cons:
- issues described in this thread
I'm about to submit my thesis next week but I can get to this after that. sorry about the issues @js2264 !