CLN: remove PY2

[jreback]

master issue for removing all vestigaes of Python 2.

After merging: #24942 (removing the PY2 CI jobs) and #25554 (bump numpy to 1.13.3), we will be ready to remove PY2 code, tests, and doc references.

Let's do this in bite sized pieces. IOW don't do a PR with changes for everything.

cc @pandas-dev/pandas-core

[TomAugspurger]

Have we discussed whether we continue to support 3.5? Matplotlib master seems to be on 3.6.

[WillAyd]

I think it would be great to jump to 3.6 eventually for better typing support, f-strings, dict ordering, etc... but at the same time I think it makes sense to rip out the Py27 chunks first before going that route, as there will be plenty as is to clean up

[jreback]

step by step here. Let's remove PY2 first. Agree not against dropping 3.5 but should not do all at once.

[TomAugspurger]

CF seems to have deprecated 3.5 building? http://conda-forge.org/docs/user/introduction.html#pinned-announcements Anaconda has stopped building 3.5 packages.

…

On Thu, Mar 14, 2019 at 10:30 AM William Ayd ***@***.***> wrote: I think it would be great to jump to 3.6 eventually for better typing support, f-strings, dict ordering, etc... but at the same time I think it makes sense to rip out the Py27 chunks first before going that route, as there will be plenty as is to clean up — You are receiving this because you are on a team that was mentioned. Reply to this email directly, view it on GitHub <#25725 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/ABQHIiP4yYwqpjHFLSANkOOFUXYO5vgXks5vWmsagaJpZM4b0Bb6> .

[TomAugspurger]

That's actually affecting @h-vetinari's other PR; pytest wasn't being bumped correctly on the 3.5 build (since Anaconda doesn't have new enough 3.5 packages for it).

[jreback]

ok merged the removal of the CI jobs for PY2: #24942

open season on PY2 removals. but let's do this in a measured way.

cc @pandas-dev/pandas-core

[h-vetinari]

Before it's open season, I'd suggest to at least merge #25752 first - currently the CI is much reduced from what it was due to the requested hard deletion of PY2-jobs in #24942 (instead of replacing them with new CI jobs).

Another possible candidate to wait for is #25554 (which depends on #25752 in some sense).