Remove vendored typing-extensions #13336
Conversation
ichard26
added
type: refactor
src/pip/_internal/utils/retry.py
Outdated
|
|
||
| if TYPE_CHECKING: | ||
| # TODO: import from typing once Python 3.10 is dropped. | ||
| from pip._vendor.typing_extensions import ParamSpec |
There was a problem hiding this comment.
I think this should import directly from typing_extensions and then add it's to the mypy pre commit additional dependencies:
Line 36 in ab40476
There was a problem hiding this comment.
Yeah, I need to fix the vendoring imports. vendoring is complaining because pip._vendor.typing_extensions does not exist anymore. I'll do that once Rich accepts my PR and cuts a new release.
|
Please be aware of #13313. I have been holding off on merging it, so that I can include any last minute upgrades to our dependencies before the release, but I don't want to get in a merge mess with this. If you're hoping for this to go into 25.1, let me know and I can merge #13313, to give you a stable basis to work from. |
|
Ah, I just noticed this relies on the rich PR, so I imagine it won't make it into 25.1 - that would need the rich PR to land, and a new release of rich, all within a week. |
Yup :) |
|
I completely missed this and raised my own rich PR which has now landed Textualize/rich#3763. I'm going to add this to the 25.2 milestone to remind us, can be removed if rich doesn't make a release (I imagine their releases will be less frequent with the business side of textualize having wound down). |
|
It's clear that rich is effectively stable and not going to see a release anytime soon. I'm removing this from 25.2 milestone. |
How about labelling it with |
I don't think anyone is monitoring at random labels on pip. I however am keeping tabs on rich releases for my professional job, so I'll ping back here on a new release. |
What makes you say that? It's had a number of commits about a month ago, and 4 months since the last release isn't that long. The necessary PR was merged in June. It feels a bit harsh to write rich off just because the commercial side of the development has wound down. Yes, we may not see a release in time for pip 25.2, so I'm fine with removing it from the 25.2 milestone, but why not just move it to the 25.3 milestone and check in when that comes up? |
|
@pfmoore there's been some recent developments with the maintainer: https://textual.textualize.io/blog/2025/05/07/the-future-of-textualize/ I've seen comments on GitHub issues that they're not taking on any new features that they think could be implemented outside the library. I expect release pace to slow. |
|
I was aware of that, but there's the final comment in that post:
It's not as if we're waiting on a new feature, or even a PR to be merged. We just need a release, and I don't see any indication that they've stopped doing releases (although the need for releases will have slowed). Anyway, we can wait and see. IMO, keeping this in our upcoming release milestone keeps it on the radar - the saving we get justifies that, IMO. We can just do a quick ckeck each release, and push this forward if there's no change. |
|
I just forgot to add it to the 25.3 milestone. And yes, I regret my tone. It did come off more harshly than I intended. I just don't think this PR is worth my time given rich doesn't has any plans to cut a release soon. |
|
Rich v14.1.0 has just released FYI: https://github.com/Textualize/rich/releases/tag/v14.1.0 |
ichard26
force-pushed
the
removal/typing-extensions
branch
from
770c092 to
f68454b
Compare
ichard26
force-pushed
the
removal/typing-extensions
branch
from
f68454b to
4af2afe
Compare
|
@pfmoore If you have time for a review, I'd be happy to include this in the 25.2 release. It should be a relatively straight-forward change. Feel free to say no though! I understand you're tight on free time 🙂 |
It's a shockingly large dependency that does effectively nothing at runtime for us.
Depends on Textualize/rich#3700.