Revert changes to comparison between Series[datetime64ns] and datetime.date

[TomAugspurger]

Revert the changes in #21361

• [ ]: Remove check in core/ops.py
• [ ]: Remove test pandas/tests/tseries/test_arithmetic.py::TestTimestampSeriesComparison.test_dt64_ser_cmp_date_warning
• [ ]: Reenable test pandas/tests/tseries/test_arithmetic.py::TestTimestampSeriesComparison.test_dt64ser_cmp_date_invalid

Target this for 0.24?

xref #21152

[TomAugspurger]

@jreback @jorisvandenbossche can we do this for 0.24.0? That would mean

• <0.22 : compare equal
• 0.23.0 : compare unequal
• >=0.23.1,<0.24.0 : compare equal & warn
• 0.24.0 : compare unequal

[jreback]

that looks good to me

[jorisvandenbossche]

We typically have deprecations for longer than a single release. Is there a good reason to not do that here?
(it's not that the current situation is actively harmful or so)

[TomAugspurger]

If this were the behavior in 0.23.0, I would agree with a longer deprecation cycle. Should the fact that 0.23.0 and 0.23.1 behave differently change our thoughts at all? (I'm not sure really)

[jorisvandenbossche]

I don't think that that is a reason to have a shorter deprecation cycle (at least it does not give any problems with the different behaviour, as if you already changed your code to the correct way when using 0.23.0, this will also work for the later versions with the reverted comparison op)

[TomAugspurger]

That's fair.

…

On Fri, Jun 8, 2018 at 10:59 AM, Joris Van den Bossche < ***@***.***> wrote: I don't think that that is a reason to have a shorter deprecation cycle (at least it does not give any problems with the different behaviour, as if you already changed your code to the correct way when using 0.23.0, this will also work for the later versions with the reverted comparison op) — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub <#21359 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/ABQHIlBRaAra-hbD9FiTCLQ1wmZhaB_rks5t6p9KgaJpZM4Ue0H-> .

[jbrockmendel]

#21359 (comment) looks like the plan was to implement the new behavior in 0.24.0. Any objection to doing it now?