Adding Pants build tool about python3statement.github.io

[Carreau]

Hi there,

Sure of course we are interested, as it is a good signal for users that Python 2 support is comming to an end.

Have a look at the practicality section once you want to drop Python 2. It contain useful information to keep things working for Python 2 users. Mostly it tell you how to ensure pip install the latest Python 2 compatible version for Python2 users without having to pin things, and explain the failure modes. When you do that feel free to ask for an extra pair of eyes here to make sure all have been done correctly.

Thanks !

[Eric-Arellano]

Great! Two potential blockers for adding:

1. We're about two weeks away from finishing the port to Python 3. I don't know if you require the library to already be ported.
2. We haven't committed to a firm deadline with dates, beyond knowing that we plan to drop within two major releases, which should be well before Jan 2020. I see there is both the list and the timeline, so maybe this is enough for the list?

Let me know if either needs to be addressed before submitting a PR.

[takluyver]

That's ok: it's not necessary to have already ported it, or to have a firm release date. It's sufficient if the developers of the project are on board with an intention to require python 3 by 2020.

…

On Thu, 26 Jul 2018, 1:51 a.m. Eric Arellano, ***@***.***> wrote: Great! Two potential blockers for adding: 1. We're about two weeks away from finishing the port to Python 3. I don't know if you require the library to already be ported. 2. We haven't committed to a firm deadline with dates, beyond knowing that we plan to drop within two major releases, which should be well before Jan 2020. I see there is both the list and the timeline, so maybe this is enough for the list? Let me know if either needs to be addressed before submitting a PR. — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub <#139 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AAUA9VoX6v5hZUF-pYfnVWChYaVld8nCks5uKRKYgaJpZM4VfXzq> .

[benjyw]

I'd like to clarify something here: We are talking about Pants itself requiring py3 to run, and utilizing py3 features in its own codebase.

But we intend it to be able to build py2 code (while itself running on py3) into the far future, certainly beyond Jan 2020.

So,

A) I don't know if this meets the intent of this statement.

and

B) I'm not sure if Pants joining this statement might send the wrong signal to users, that we will be dropping support for building their py2 code.

[Eric-Arellano]

I'm closing this out. We discussed offline that as @benjyw brings up, Pants is probably not the right fit for this statement.

Whereas most libraries are directly called and are going to require Python 3 to work, Pants will for a very long time support running Python 2. The tool itself will be written in Python 3, and if users want to use it as a library (a valid use case), they will need Python 3.

But to simply use the tool, Python 2 will still work. So, the statement would create confusion.

My apologies to all for jumping the gun on this!

[takluyver]

No problem. I think that's a reasonable objection for a range of developer tools that run as command line tools rather than Python libraries. Thanks for opening the discussion!