Twine 2.0?

[bhrutledge]

There's at least a few open PR's that seem like good candidates for a major release with potentially breaking changes, and it seems like there's some interest in making that release. I'm curious what other @pypa/twine-maintainers think.

• Improve the check command output - Pull Request #488
• Make twine Python 3 only - Pull Request #437
• WIP: mypy via monketype - Pull Request #469

Also related: Prepare twine for 1.14.0 release - Pull Request #468

[jaraco]

My usual inclination is not to bundle backward incompatibilities into releases.... as it makes it even more difficulty to separate those concerns. Better would be to have lots of incompatible releases, twine 2, 3, 4 each with a separate incompatibility. Then someone can accept all (with twine 4) or only some or none based on which version they elect.

However, this strategy relies on the ability to easily and rapidly make releases, which twine doesn't (yet) have #493.

In any case, I'm happy to make backward incompatible releases with any number of incompatibilities as long as the users have a clear indication of how to respond to those changes.

[pradyunsg]

Honestly, only #437 has any significant user-side consequences. And well, the only actionable thing for them is to migrate away from Python 2, which is perfectly fine.

In its current state, #469 gets unblocked by #437 and will not have any user-side consequences. Introducing static typing to a codebase does not require any additional end-user system requirements. Even depending on typing being available at runtime isn't necessary but it's OK if you're on Python 3 only since it's in the standard library.
(related: https://github.com/pypa/pip/blob/master/src/pip/_internal/utils/typing.py#L12-L14)

#488 is an improvement to the check command output. I've noted my thoughts on treating that as "backward incompatible" here: #488 (comment). tl;dr -- don't, because you'll have to treat every change to twine's output as backward incompatible, but it's call for y'all to make and I'm OK with whatever gets this done quicker.

[jaraco]

I've released 1.15 with #488.

[jaraco]

In #498, I propose releasing 2.0 with the Python 3-only change (and one other). I'm happy to roll out #469 when it's ready.