Python 3.6.0rc2 coming soon, 3.6.0 final now 2016-12-23
Hi all!
Today (2016-12-16) has long been our planned release date for 3.6.0 final. So far most of the feedback from users testing the preview versions of 3.6.0 has been very positive. We made it to the next-to-final milestone, the 3.6.0rc1 release candidate, 10 days ago with hopes of going directly to the final release. Alas, in the last few days at least one outstanding issue that we had hoped would not be a real-world problem has proven to be a showstopper during third-party package testing and I have been persuaded that we do need to fix it before 3.6.0 final. I take responsibility and apologize for not ensuring it was resolved earlier in the release cycle; I'll try to do better next time. Therefore, we are going to produce a second release candidate. Besides the showstopper fix (#28147), I've cherrypicked a few requested build fixes and, as promised, some last-minute documentation updates and additions. I'm also expecting to cherrypick at least one more asynchio fix before tagging and manufacturing 3.6.0rc2 sometime later today (i.e. within the next 24 hours). Assuming that is accomplished, we will be looking for quick feedback to ensure that we have addressed the problems and have not introduced any new ones. Then, assuming all goes well and no new showstoppers are found, we plan to release 3.6.0 final on Friday 2016-12-23, a week from now.
Also note that there is no change in the status of the cpython repo branches. Continue to push appropriate changes to the 3.6 branch for the 3.6.1 maintenance release and to the default branch for the next feature release, 3.7.0. Should you run into a potential showstopper problem for 3.6.0, please make sure there is an open issue for it on the bug tracker marked as "release blocker", work to getting a fix pushed to the 3.6 branch for 3.6.1, and contact me ASAP to discuss potential cherrypicking. Please do the same for any necessary important documentation fixes for 3.6.0 final. As before, my goal will be to have no new changes after the release candidate.
Thank you all again for your great efforts and co-operation throughout the 3.6 development cycle! We are oh-so-close to getting your work officially out there.
--Ned
P.S. Happy Beethoven's Birthday
FYI: Here is a list of the post 3.6.0rc1 changesets that have been cherrypicked so far for 3.6.0rc2. There will likely be at least one more. (Note, the description and files list below for some changesets may be truncated.)
user: Yury Selivanov
Ned Deily:
Alas, in the last few days at least one outstanding issue that we had hoped would not be a real-world problem has proven to be a showstopper during third-party package testing and I have been persuaded that we do need to fix it before 3.6.0 final. I take responsibility and apologize for not ensuring it was resolved earlier in the release cycle; I'll try to do better next time.
No need to apologize, you are doing a great job! I'm not surprised at all that major bugs are still found just a few days before the final release: many people wait just before a final release to test their code. I'm happy that such bugs are found _before_ a release. Bugs like "my applications takes 4 GB of memory with Python 3.6 but 40 MB with Python 3.5" (#28147) seem so big that it would be a shame to "ship" such bug in a final release!
FYI: Here is a list of the post 3.6.0rc1 changesets that have been cherrypicked so far for 3.6.0rc2. There will likely be at least one more. (Note, the description and files list below for some changesets may be truncated.)
test_gdb is broken in the RC1. To fix test_gdb, I convinced Ned to
also cherry-pick:
---
changeset: 105342:752863f96fb8
user: Victor Stinner
participants (2)
-
Ned Deily
-
Victor Stinner