docs: prepare for 2.10.1 release #4279
Conversation
Signed-off-by: Henry Schreiner <[email protected]>
There was a problem hiding this comment.
To the other maintainers: please vote your mind.
I feel digging into the suspected ODR question is most important at the moment, expecting it's only "hours" (ballpark) of effort to get more clarity, not "days". So that's what I'll do now.
I'm not opposed to this release anymore after coming to believe that the internals bump won't help, but I still feel more clarity is needed.
docs/changelog.rst
Outdated
|
|
||
| Changes: | ||
| Using at least this version is recommended if you need Python 3.11 support |
There was a problem hiding this comment.
Suggested change:
This release has one change (PR #4119) that improves compatibility with Python 3.11 for users of pybind11/embed.h.
I think it's important to highlight, but I'd let people decide by themselves how important this is to them. (I assume it's not important to most users.)
There was a problem hiding this comment.
Reworded to clarify that this the first version to support embedding 3.11.
|
@henryiii I could make the release tomorrow, I think I remember enough + the documentation that I can figure out all the steps. Please let me know if that would be helpful. |
|
That would be, I won't have a consistent amount of free time to make the release in the next few days. Changes above look fine, and we ideally want one more PR (the GIL_whatever_we_called_it) one. And maybe your error export one, though I should point out that's not an issue reported by users, so I'd be careful about making it too critical - if it's a largish change, it needs to sit in master a bit. |
Thanks! I'll take care of it.
Will take care of it.
I looked at the list of my open PRs but it didn't really click: which PR? This one? #4072 (comment) Just the visibility fixes, but not the cmake change? (I was thinking myself it's best to split the PR.) |
|
#4072 needs to wait and get reviewed properly, and is not a target for 2.10.x. I just meant whatever you are working on for the ODR worries in follow up to the ABI issue, if you think it should go into 2.10. |
Oh good. I think "trimming back the scope #2999" (to reduce the scope for ODR worries) does not have to go into 2.10. So just this PR and #4216, then we're good. |
|
Waiting on, and including, #4297. |
Signed-off-by: Henry Schreiner <[email protected]>
|
I changed (simplified) the PR description, in case someone looks here later. The previous version was outdated. |
* docs: prepare for 2.10.1 release Signed-off-by: Henry Schreiner <[email protected]> * Update changelog.rst * docs: update changelog with final list of PRs Signed-off-by: Henry Schreiner <[email protected]> * Update docs/changelog.rst * chore: one more changelog bump Signed-off-by: Henry Schreiner <[email protected]>
* docs: prepare for 2.10.1 release Signed-off-by: Henry Schreiner <[email protected]> * Update changelog.rst * docs: update changelog with final list of PRs Signed-off-by: Henry Schreiner <[email protected]> * Update docs/changelog.rst * chore: one more changelog bump Signed-off-by: Henry Schreiner <[email protected]>
Description
Work on #4216 exposed deadlocks previously mistaken for flakes. This condition existed at least for a couple years and will be further debugged after the 2.10.1 release.
Closes #4125. CC @wjakob, @rwgk, @Skylion007, @ax3l.