Hello list,
Tomorrow I am going to do one last pass at the 0.14 milestone and then release 0.14. As a reminder, 0.14.x will be a LTS release, the last skimage version compatible with Python 2.7, and it will receive important backports (bug fixes, critical performance improvements, and maybe even a new feature or two, based principally on community demand).
Given how long we have been planning this release, I think we have reached the point where we should release what we have rather than hold it up for every last PR that we want in it. Many users have requested this release, and it is particularly important for our Windows users that may not have a complete developer toolchain to build from source, and have a difficult time accessing the features that have made it onto master in the past year.
If you have any strong objections to pushing this out, please voice them now.
Egor, you twice mentioned wanting to be release manager for this release. (Sorry that I was not more available to help on those occasions.) I’m happy to let you do this one if you still want to, but I’m also happy to do it myself. Let me know your preference.
Thanks,
Juan.
Hi Juan,
Yes, I'm still willing to be a release manager. I agree with you that, despite having a moderate list of open issues [softly] scheduled for 0.14, we should go ahead and make a release. The master branch holds a ton of fascinating additions, and it is way far ahead of 0.13.1 already.
My suggestion would be to wait couple of days so that the maintainers could review 0.14 milestone and catch up with the issues they consider critical for 0.14. Once everything is set, we (I might need your assistance, @jni :)) can make a release (this Thursday or Friday).
Regards, Egor
пн, 21 мая 2018 г. в 15:57, Juan Nunez-Iglesias jni@fastmail.com:
Hello list,
Tomorrow I am going to do one last pass at the 0.14 milestone and then release 0.14. As a reminder, 0.14.x will be a LTS release, the last skimage version compatible with Python 2.7, and it will receive important backports (bug fixes, critical performance improvements, and maybe even a new feature or two, based principally on community demand).
Given how long we have been planning this release, I think we have reached the point where we should release what we have rather than hold it up for every last PR that we want in it. Many users have requested this release, and it is particularly important for our Windows users that may not have a complete developer toolchain to build from source, and have a difficult time accessing the features that have made it onto master in the past year.
If you have any strong objections to pushing this out, please voice them now.
Egor, you twice mentioned wanting to be release manager for this release. (Sorry that I was not more available to help on those occasions.) I’m happy to let you do this one if you still want to, but I’m also happy to do it myself. Let me know your preference.
Thanks,
Juan.
scikit-image mailing list scikit-image@python.org https://mail.python.org/mailman/listinfo/scikit-image
Hi Juan,
On Mon, 21 May 2018 22:48:15 +1000, Juan Nunez-Iglesias wrote:
Tomorrow I am going to do one last pass at the 0.14 milestone and then release 0.14. As a reminder, 0.14.x will be a LTS release, the last skimage version compatible with Python 2.7, and it will receive important backports (bug fixes, critical performance improvements, and maybe even a new feature or two, based principally on community demand).
Thanks for pushing ahead with this release!
One request: that we add one review step to releases, which is that we (as a community) review the release notes. In the past, the release manager simply did these by themselves, but since it is so widely visible, additional eyes would help (to be clear, I'm referring mainly to my own past attempts to get this right!).
Alternatively, we can try doing a pre-release, like NumPy and SciPy do. That is probably the safer and more responsible option, since we may also hear from other projects that have the "--pre" flag in their continuous integration setups. But it happens to also give us a chance to complete the above-mentioned release note review.
Josh, we can still backport flood fill to a next point release later, especially if you can already submit it by tomorrow.
Best regards, Stéfan
Hello everyone, and thanks for the robust response! =)
Josh, there is no urgency in getting flood fill into 0.14.0. It can go into 0.14.1 equally well. Actually my goal is for 0.14.0 to be out before next week's sprint *and* possibly to do generic flood-fill with LowLevelCallable support during the sprint. These goals are mutually exclusive with having flood fill in 0.14.0. And, incidentally, I would think that nD basic flood-fill should be as easy as 2D given the offsets_to_neighbors functions in watershed/the local_maxima PR. I would *very* much prefer to have an nD flood-fill from the beginning. What is "basic" flood fill btw? Same value as starting index?
Egor, awesome that you can do the release. As to the timing, given Stéfan's suggestion that we should have a round of review of the release notes before the actual release, *and* that I am flying MEL-SFO on Thursday, do you think we could do this today or tomorrow (Wednesday)? Today would be much less hectic for me, but tomorrow would be ok. What is your timezone btw?
Stéfan, as to pre-releases, it is a good idea, but I would target that for 0.15 or maybe even 0.14.1. I don't want to change the release process right this second.
Onwards and upwards!
Juan.
-
Egor Panfilov -
Juan Nunez-Iglesias -
Stefan van der Walt