use github actions instead of travis-ci #968
Conversation
|
#967 took close to 2 hours to return results, let's see if github actions can do better (and test windows too!) |
|
oh hmmm, looks like actions might need to be enabled before I can try this 🤔 |
|
@IanLee1521 or @sigmavirus24 can you enable github actions on this repository? |
|
Unless this a race condition with @sigmavirus24 and they just beat me to it; it looks like Actions are enabled (below screenshot). Maybe Travis CI has to finish running first?
|
|
Hmm, I'm not sure that Travis is running... maybe because this is a WIP pull request? |
|
Also, I think actions need to be on master for them to run |
|
hmmm I've submitted PRs for actions before and they run immediately, even when I don't have commit bit -- I'll play around with this in my fork (it might be easier for me to control there?) |
|
had a few mistakes but it passes now on my fork: https://github.com/asottile/pycodestyle/runs/1359541910 |
That's never been my experience, frankly. I've always seen exactly this behaviour |
| on: | ||
| pull_request: | ||
| push: | ||
| branches: [master, 'test-me-*'] |
There was a problem hiding this comment.
Probably don't want to keep 'test-me-* in here
There was a problem hiding this comment.
oh, I usually find this useful to debug CI without having to open a PR
There was a problem hiding this comment.
Stupid question: If you can push to a repo and you trust others who can push to the repo, why bother restricting this at all?
There was a problem hiding this comment.
it's mostly so when pushing branches pushed to pycqa/pycodestyle and creating a pull request based on them you don't run the actions twice
There was a problem hiding this comment.
🤷 You're still running them twice, just once on a random push, and a second time when opening the PR. They're just separate.
There was a problem hiding this comment.
🤷 I don't personally see any problems with running tests that often. But also CI systems never actually test against a merged build, just against the head of the branch so they're usually trash either way
There was a problem hiding this comment.
GA tests against the merged revision
There was a problem hiding this comment.
It doesn't update when master moves forward in my testing though so it's a point-in-time merged revision, not one before you press merge
There was a problem hiding this comment.
Not an argument against GA just a general gripe
weird, a PR worked fine for me when I was adding a wheeling job here: hynek/argon2-cffi#70 |
No description provided.