Closed yubiuser closed 1 year ago
Before I was the one who made this so for the same reasons but vice versa:
But I actually like your point about energy consumption.
What do others think?
Another option could be to run the style check first, but add a continue-on-error: true
for this part. This way, PRs still get checked thoroughly if styling is wrong, but formatting feedback is given as fast as possible.
However, this would defeat the benefit of less energy consumption.
Personally I think people should get into the habit of running the linter(s) locally. That would give the fastest feedback and prevent the uninteresting failures propagating here for all to see.
Sure, always running linters locally would be the best solution. But until this is done by everyone you might still want to check during the workflow. I don't think running locally and running during the workflow are mutually exclusive.
Correct, not mutually exclusive, I didn't mean to imply that.
So what do we do? :smile:
It's the burden of the maintainer to decide. But I won't be resentful if the PR is rejected ;-)
So I'll go ahead and merge this one, we can always change later if we want to.
Running the test workflow is a time and resource consuming task - the workflow runs between 20 and 40 minutes. Currently, the heavy test (compiling and cross-compiling) run before the lighter linting and format checking. This follows the following reasoning:
However, I think the order should be changed to check formatting and linting first: