Conversation
In `PyFilePlugin.__init__()`, the `filename` must be wrapped in a list before passing it to `importlib.util.spec_from_file_location()`, whose `submodule_search_locations` argument is expected to be a list of `str`. That +1/-1 patch should be released as a hotfix for 8.0.x, but the rest of the type changes and doc tweaks can wait for 8.1.
Exirel
left a comment
There was a problem hiding this comment.
I really really like that. Only a nitpick.
| # core | ||
| self._regexes = regexes | ||
| self._plugin_name = plugin | ||
| self._plugin_name = plugin or '' |
There was a problem hiding this comment.
Given plugin is always a str now, there is no need for that (unless mypy complain? but that sounds weird):
| self._plugin_name = plugin or '' | |
| self._plugin_name = plugin |
Note: I'm not even sure why I made plugin optional in the first place.
|
Hmm, I'm torn whether to resolve the conflicts on this. It's doing some unique stuff, but perhaps it's better to wait until after #2669 is done mucking about in |
Oh, yeah, that's probably better. Sorry. 🥲 |
|
@dgw so many conflicts now... 😭 |
|
Oof, when I get back to this I might reset the branch and start over instead of going through all 40-something conflicts. I bet that would be easier. |
Description
This does what it says on the tin, and is intended to land in 8.1.0.
Starting as a draft, because it has existing merge conflicts already (I think they hail from #2616, but haven't checked
git blamefor all of them) and also includes part of #2633 (which needs to land in 8.0.1 instead).Checklist
make qa(runsmake lintandmake test)Notes
I probably didn't open a PR from this branch last month because I wasn't sure it was done, but now that a good 4 weeks went by and seeing this in the list of branches made me think, "Wait, what's that for?" I think I'd better get in the habit of opening draft PRs sooner rather than later. #2633 was another forgotten branch that I thought had been turned into a PR but hadn't yet.