You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
All our CI jobs run at the same time but some are quite long (e.g R CMD check because of #37) while others should be quite quick (e.g styling, linting, and spellcheck).
The problem is that these quick checks might require tiny changes that don't influence the code at all, like fixing a typo or deleting a new line. When this happens, the new commits re-trigger the whole suite of tests, even R CMD check which shouldn't be affected by these tiny changes. Therefore, maybe there should be a "first run" of small CI jobs that check that everything's fine and when this passes, then we trigger the longer jobs like R CMD check.
I don't know the specific name of this but it looks like this in GHA (example from r-universe workflows):
I'm not sure how this would influence our dev workflow but it just annoys me to lose a few minutes every time there's a lint or some style to fix (and it also wastes some GHA minutes or just computing power). Would something like this make sense to you?
The text was updated successfully, but these errors were encountered:
Note that for those tiny checks, we could use ubuntu-latest with r2u to have a faster setup (currently lint-changed-files takes about 3 min to setup R and install the deps)
All our CI jobs run at the same time but some are quite long (e.g R CMD check because of #37) while others should be quite quick (e.g styling, linting, and spellcheck).
The problem is that these quick checks might require tiny changes that don't influence the code at all, like fixing a typo or deleting a new line. When this happens, the new commits re-trigger the whole suite of tests, even R CMD check which shouldn't be affected by these tiny changes. Therefore, maybe there should be a "first run" of small CI jobs that check that everything's fine and when this passes, then we trigger the longer jobs like R CMD check.
I don't know the specific name of this but it looks like this in GHA (example from r-universe workflows):
I'm not sure how this would influence our dev workflow but it just annoys me to lose a few minutes every time there's a lint or some style to fix (and it also wastes some GHA minutes or just computing power). Would something like this make sense to you?
The text was updated successfully, but these errors were encountered: