maint: use pytest-cov, cleanup docs/requirements.txt, and reduce rate limiting issues in CI#169
Conversation
It becomes clear that this isn't a repo configured secret, but the token made available by github's system.
consideRatio
changed the title
| "myst_parser", | ||
| "sphinx_book_theme", | ||
| "sphinx>=5", | ||
| "sphinx-design", |
There was a problem hiding this comment.
this was part of docs/requirements.txt, but not here, so I added it here.
| # Use TOKEN_READONLY if available (pushed branches), otherwise use GITHUB_TOKEN (PRs) | ||
| # TOKEN_READONLY is a PAT for @choldgraf that only has read-access to this repo but isn't available in PRs. | ||
| GITHUB_ACCESS_TOKEN: "${{ secrets.TOKEN_READONLY || secrets.GITHUB_TOKEN }}" |
There was a problem hiding this comment.
Use of this token failed in another way then using the workflow's github.token (aka secrets.GITHUB_TOKEN), so instead of debugging it etc, I removed it and reduced the burden on rate limiting to stabilize things that way.
|
Got time for more review @nabobalis? ❤️ 🎉 I'd also love to help do a release with #166, and if so, do we use version 1.1.6? I added a bug label to it. |
choldgraf
left a comment
choldgraf
left a comment
There was a problem hiding this comment.
This all looks pretty clean to me! So you're reducing the number of parallel builds in order to lessen the API hits, and removing my read-only token in the process. I think that's safe to try and maybe we can optimize further if we start failing CI because we hit the rate limit.
|
Use of your CI token failed in some other way, i didnt look into and figure out why, but perhaps it didnt have all the required permissions when it started to inspect github releases or something. |
|
Thank you both for helping out with reviews @choldgraf and @nabobalis! |
|
To be clear I would love to not use my tokens at all haha so I'm a big fan of this. I added it a few months back because I was getting horrifically frustrated with our CICD haha |
|
All green in main :D |
|
omg quick don't touch anything 😅 |
2 participants
AI disclosure
codexand its GPT 5.5 model was used to determine the state of things, and wrote the concurrency / max-parallel parts in the CI. Inline comments are my own, and everything else.