RLS: 0.24.x

[jreback]

Tracking issue for 0.24.1

[jreback]

@pandas-dev/pandas-core so it is very important to tag before you merge to a backport branch, otherwise the auto-backport-bot won't be triggered, e.g. we want to see something like #24951. Please make sure that these pass and then merge them (or you will need to manually fixup / merge to the backport branch).

[jorisvandenbossche]

I created the branch, so from now on, if you label a PR with 0.24.1 before merging, a backport PR will be created automatically if possible (and if not possible, a label will be added).

[jreback]

@jorisvandenbossche great thanks

[jorisvandenbossche]

And since Jeff and I already had some discussion about it :)

IMO, we should not be backporting too much, but limit ourselves to regressions and important bug fixes.

Backporting any tiny doc fix / bug fix is simply a) more work, and b) gives a higher risk of also breaking something unneeded.

[TomAugspurger]

Thought on doing a 0.24.1 soonish? Tracking a few regressions

• df.to_dict('records') broken when 255+ columns present #24939
• Index.intersection changed behavior to sort by default in pandas 0.24 #24959

I'd like to include #24964 as well (probably not a regression in 0.24, but I think important to have working, since that's hurting our import times).

any others?

[jreback]

we need to wait a few weeks - not worth it to do for these minor things

[TomAugspurger]

FYI releases are fairly cheap to do these days.

…

________________________________ From: Jeff Reback <notifications@github.com> Sent: Sunday, January 27, 2019 16:39 To: pandas-dev/pandas Cc: Tom Augspurger; Team mention Subject: Re: [pandas-dev/pandas] RLS: 0.24.1 (#24949) we need to wait a few weeks - not worth it to do for these minor things — You are receiving this because you are on a team that was mentioned. Reply to this email directly, view it on GitHub<#24949 (comment)>, or mute the thread<https://github.com/notifications/unsubscribe-auth/ABQHItJgmvDciEWKS6uB8hb51s4-LvgYks5vHiqqgaJpZM4aUMuQ>.

[h-vetinari]

+0.5 for a fast release to fix the regressions before they got much chance to be relied on.

[jorisvandenbossche]

+1 on doing quick releases with regression fixes, if it is not a lot of work.
Certainly for the intersection sort behaviour it would be good to do this as fast as possible

[TomAugspurger]

I'd like to do a release in the next day or two. What are peoples thoughts? (I'll be busy next week, and unable to do the release myself. Others could, but I think getting a release out soon with the fixes for iteritems and the index setops is important).

Right now we have a decent number of issues on the 0.24.1 milestone, which I'm going through.

[jorisvandenbossche]

I think only the regressions + the setops default change that are in the 0.24.1 milestone are important. +1 to do a release shortly.

[TomAugspurger]

I'd like to do a release tonight if possible, with the following issues resolved

• Index.intersection changed behavior to sort by default in pandas 0.24 #24959
• Regression in DataFrame.set_index with class instance column keys #24969

Any objections? Any other blockers?

[TomAugspurger]

Tagging in about 5 hours from now.

[TomAugspurger]

Waiting on #25085 to pass CI, then merging, back porting that, and tagging.

[jorisvandenbossche]

@TomAugspurger Thanks a lot for the release!

One thing I noticed: due to the way we do the backporting now, the actual contributors don't get acknowledged: http://pandas.pydata.org/pandas-docs/version/0.24.1/whatsnew/v0.24.1.html. That section is quite useless/misleading now, so I would either update it manually or remove it.
In general, I think it could be solved by the bot using co-authors: scientific-python/MeeseeksDev#35

[jorisvandenbossche]

(repurposed this issue for a potential next 0.24.2)

[jorisvandenbossche]

We already have a nice set of changes in 0.24.x, so shall we try to do a 0.24.2 release this week?

I have a little bit of time to look at some of the remaining open issues, but not that much. @TomAugspurger would you have a bandwidth to do a release? Otherwise I might be able to do early next week.

For the remaining open issues, I would certainly try to fix #25193 (the pip install problem), and some of the others seem to have a relatively straightforward fix (eg #25077, #25403).

[TomAugspurger]

I likely won't have time this week. Possibly next week if needed (though it'd be good to exercise the release scripts on a machine other than my own :).

…

On Mon, Mar 4, 2019 at 9:17 AM Joris Van den Bossche < ***@***.***> wrote: We already have a nice set of changes in 0.24.x, so shall we try to do a 0.24.2 release this week? I have a little bit of time to look at some of the remaining open issues, but not that much. @TomAugspurger <https://github.com/TomAugspurger> would you have a bandwidth to do a release? Otherwise I might be able to do early next week. For the remaining open issues, I would certainly try to fix #25193 <#25193> (the pip install problem), and some of the others seem to have a relatively straightforward fix (eg #25077 <#25077>, #25403 <#25403>). — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#24949 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/ABQHIoNuqPnHAHhyrSjE6e0BB-NpJzCSks5vTTkQgaJpZM4aUMuQ> .

[jorisvandenbossche]

Will be tagging 0.24.2 this afternoon (trying out the release scripts now)

[jorisvandenbossche]

MacWheels tests are failing: https://travis-ci.org/MacPython/pandas-wheels/builds/505474702 (on python 3.6 a stata failure, and on python 2.7 numpy ufunc tests).
I am in a workshop during the day, so if somebody has time to look at it, certainly welcome (otherwise I will look at it this evening).

[simonjayhawkins]

on python 2.7 numpy ufunc tests

#25186 fixes

[TomAugspurger]

Picking up the remaining release items now.

TODO:

• Fix MacPython wheels (fix MacPython pandas-wheels failure #25186)
• Upload to PyPI
• Build & upload pdf docs
• symlink 0.24 -> 0.24.2
• update website
• Announce

Other

• Backport fix MacPython pandas-wheels failure #25186 to 0.24.x

[simonjayhawkins]

@TomAugspurger #25186 only for the numpy ufunc. there is still the stata issue.

[TomAugspurger]

Thanks. Still investigating, but it's possibly a flaky test: #25707

[jorisvandenbossche]

All wheels are uploaded, and website symlink is changed.