Prepare release notes for v0.20.0#6556
Conversation
produced by generate_release_notes.py. Still requires manual cleaning.
"New features" MUST be totally backwards compatible. Everything else goes into "changes and new deprecations".
stefanv
left a comment
There was a problem hiding this comment.
Thanks for starting this.
The biggest issue I see is that some PRs are listed twice.
|
@lagru Thanks for starting this. I was just chatting w/ Stefan about the 0.20 release. I tentatively volunteered to make the 0.20 release and am happy to do it, if it is helpful. According to the schedule, Python 3.11.0 final is due Monday, 2022-10-24. I would like to release 0.20 just before or just after the 3.11 release with 3.11 wheels. NumPy, matplotlib, Pandas, and a few other packages have already released 3.11 wheels. My preference (for what it is worth) would be to release 0.20 before 10/24 and move all the remaining issues and PRs for 0.20 to 0.21. If we are going to release 0.20 before 10/24, we should release a 0.20rc1 10/17 or before. The main thing left that I would like to see merged first is the meson stuff and I would like to bump the NetworkX requirements. For NetworkX, we are about to release version 3.0 and it would be good to have scikit-image require at least 2.8 as a minimum NetworkX requirement as it has all the pending deprecations for NX 3.0. |
|
Thanks everyone. I'll go through the suggested changes. A lot of the stuff I just haven't gotten to yet. @jarrodmillman I don't have a preference between your suggested release dates. I'll try to get this is shape bit by bit until the middle of next week if that's early enough. |
Some suggestions I plan to edit locally, some I'd like to discuss first. Co-authored-by: Stefan van der Walt <sjvdwalt@gmail.com> Co-authored-by: Marianne Corvellec <marianne.corvellec@ens-lyon.org>
Use present tense and imperative tone for release notes. Try to be more precise about why and how things changed. Co-authored-by: Marianne Corvellec <marianne.corvellec@ens-lyon.org> Co-authored-by: Stefan van der Walt <sjvdwalt@gmail.com>
A lot of the removed PRs were already backported to 0.19. So I'm ignoring these for 0.20. Note that I'm checking if the commit is available in these releases with "git tag --contains <COMMIT_HASH>".
0e76df3 to
122c3fa
Compare
I'm confused with what to do abou the Milestone 1.0. Should these be included in 0.20 now. Collect them for now until that's confirmed. Furthermore, collect the backported PRs. We can delete them later but for now it's useful to keep them for review and in case we want to go back. Note that quite a lot are missing because I didn't start this earlier.
|
Man there's some nice things in here! 😊 thanks @lagru! |
doc/release/release_0.20.rst
Outdated
| - Use the minimal required unsigned integer size in ``skimage.filters.rank_order`` which allows to operate the function with higher precision or on larger arrays. | ||
| Previously, the returned ``labels`` and and ``original_values`` were always of type uint32. | ||
| (`#6342 <https://github.com/scikit-image/scikit-image/pull/6342>`_). | ||
| - Use the minimal required unsigned integer internally in ``skimage.morphology.reconstruction`` which allows to operate the function with higher precision or on larger arrays. | ||
| Previously, int32 was used. | ||
| (`#6342 <https://github.com/scikit-image/scikit-image/pull/6342>`_). |
There was a problem hiding this comment.
There was a problem hiding this comment.
This was just referring to a place where morphology.reconstruct had unconditional promotion to float64 internally. See at the left in the diff here on lines 173-175 where seed and mask used to get copied into a float64 images array. This was changed to use _supported_float_type so it is possible to get single precision output from this function.
There was a problem hiding this comment.
Thanks for the clarification. So actually mask determines the dtype of the output and not seed?
There was a problem hiding this comment.
Good question... I would assume we would want "float_dtype = max { _supported_float_type(seed.dtype), _supported_float_type(mask.dtype) }" 🤔
Co-authored-by: Marianne Corvellec <marianne.corvellec@ens-lyon.org>
as of 2022-12-05.
Only PR left is the one with the long list of completed deprecations.
Co-authored-by: Marianne Corvellec <marianne.corvellec@ens-lyon.org>
Co-authored-by: Marianne Corvellec <marianne.corvellec@ens-lyon.org>
|
Okay, after summarizing #6583 😮💨, this should be up to date now. 😁 I'm unsure how we want to proceed. |
Thanks for all the work, @lagru! I guess that, right after merging this PR, we want to create a branch from |
I don't think we looked more at how to run/test the wheels locally, but didn't really cover all aspects of that. Given the infrequency of these releases, I always refer back to https://github.com/scikit-image/scikit-image/blob/main/RELEASE.txt We should start with tagging an alpha or release candidate, given the large number of API changes. We can then make sure the wheels seem solid before tagging the final release. I think @jarrodmillman said he may have some time to work on it after a new |
|
Forgot to mention that creating the branch does not trigger wheel builds. Only pushing a tag starting with the pattern here will trigger the wheel build action: scikit-image/.github/workflows/wheel_tests_and_release.yml Lines 1 to 6 in ae28814 Only those starting with scikit-image/.github/workflows/wheel_tests_and_release.yml Lines 225 to 236 in ae28814 |
|
Also we still have to create the highlights in first section if we want those. But that can probably be done after branching as well. |
|
@mkcor no worries. Adding an item it is easily done. ;) |
Co-authored-by: Marianne Corvellec <marianne.corvellec@ens-lyon.org>
Co-authored-by: Marianne Corvellec <marianne.corvellec@ens-lyon.org>
See https://github.com/scikit-image/scikit-image/releases/tag/v0.19.0 for an example from an earlier release.
jarrodmillman
left a comment
There was a problem hiding this comment.
We should merge this for the 0.20rc1 release and we can fix it up afterward, if needed.
|
Thanks @lagru. @jarrodmillman that sounds like a good plan.
A sentence I think we should tweak a bit to read more easily. |
Description
Start preparation of the release notes for scikit-image 0.20.0. The initial list of PRs was generated with generate_release_notes.py. Compared to 0.19 and release_template.rst I'm using different sections to structure the changes for now. This is because I'm not very clear on the distinction between suggested sections like "API Changes", "Deprecations" and "Backward Incompatible Changes".
And keep up to date...