[Pandas-dev] Just a quick question from a regular pandas user

[Joris Van den Bossche]

Hi Martin,

The 2.0 milestone is not updated for a very long time, and also not yet
really used (there are a few issues tagged with it to mean "maybe in a next
big release but not yet in 1.0"). So I wouldn't look too much to that. In
any case, we are certainly not going to do a pandas 2.0 release in summer
2020 (so we should update the milestone date).

What we *do* plan is a final 1.0 release in early 2020. What we also
discussed recently is a version policy for starting with 1.0:
https://dev.pandas.io/docs/development/policies.html#version-policy
This means that code working with 1.0 should mostly keep working in the
full 1.x series of releases when not using experimental features (although
we will keep doing deprecations, so you still might need to change code to
get rid of such warnings, in preparation of pandas 2.0).

And you are correct: pandas 1.0 will not be drastically different from
0.25.3 (the main difference will be that a lot of things that were
deprecated before will now be removed, plus some documented API changes).
While we do not yet have much concrete plans for pandas 2.0, I think the
expectation is that it will be similar (and also not something for the
coming year anyway).

So if you are writing code now for 0.25.3, and you take notice of possible
deprecation warnings and fix your code for those, you can be ensured that
your code will mostly work on 1.0 as well.
Now, it is still very recommended to ensure you write tests for your code,
so you can run those on new pandas releases to verify this is indeed the
case (and running such tests on release candidates of new pandas releases
is also very valuable, so potential regressions can be reported and fixed
early).

Hopefully that could shed some light
Joris


On Fri, 29 Nov 2019 at 05:42, Martin Gantchev <bms91 at abv.bg> wrote:

> Dear Representatives of Pandas-dev,
>
> This is Martin here, a regular user of the pandas library.
>
> First of all, thank you for providing, maintaining and still developing
> this amazing library which I use pretty much every day.
>
> On that note, I am facing a project that will involve working with pandas
> heavily, but that is supposed to retain the code for a long period of time
> (hopefully, for years to come).
>
> I am referring to this piece of information:
>
> https://github.com/pandas-dev/pandas/milestones
>
> It seems that pandas 1.0 has 90% completion rate, while pandas 2.0 is
> expected to be ready for as early as August 2020, however it strangely has
> just 10 problems that need to be solved.
>
> Of course, no precise answer is requested. However, I am afraid that in
> the next couple of months I may write code that might become obsolete in
> the middle of next summer. Am I right about that?
>
> I did read around the internet and read more articles, so I don't expect
> neither 1.0 or 2.0 to be drastically different from 0.25.3. At least, I
> guess most of the code I'd use in 0.25.3 should work normally under 1.0 or
> 2.0. Is that correct?
>
> Shedding light on this subject may save tons of worries for me, so even a
> loose delineation of your schedule and the potential impact it may have on
> code written in 0.25.3 would be greatly appreciated.
>
> Thank you very much!
>
> Looking forward to your answer.
> Best,
> Martin
> _______________________________________________
> Pandas-dev mailing list
> Pandas-dev at python.org
> https://mail.python.org/mailman/listinfo/pandas-dev
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/pandas-dev/attachments/20191129/0099f553/attachment.html>