Remove Python 2 support

[afh]

How do ~~we~~people feel about removing support for Python 2 to simplify the codebase a bit now that use of Python 2 has been deprecated for over two years?

Python 2.7 support ended on January 1, 2020, […] [the] final release, 2.7.18, occurred on April 20, 2020, […] [and] marked the end-of-life of Python 2.

(source Wikipedia)

[afh]

Thanks for your thoughts on this, @tbm.

I think keeping Python 2 support around for another release and communicating its deprecation release and/or date in some shape or form would be appreciated by those still using Python 2 support*.

This project's versioning hasn't been strictly adhering to semver—which is perfectly alright. Yet at some point I think it would be good to get an idea / clarification about future versions, especially as bumping the major version number from 3 to 4 (e.g. when there are breaking changes and removal of Python 2 support could be considered as such) ~~likely~~potentially creates confusion with Ledger 4 an experimental implementation of ledger in Haskell that is independent of this implementation in C++.

The many changes since the last release, including the added support for gpg encrypted journal files (yay!), surely suggests bumping the minor version, yet waiting for Python 2 removal in this project another 2 years until the planned 3.4.0 release date seems ill advised.

A few (hopefully helpful) questions:

• What if the next release is the last release to include Python 2 support, but will be 3.2.2 instead of 3.3.0?
• What if the next release is the last release to include Python 2 support and is 3.3.0 as planned, but the 3.4.0 release is scheduled (much) sooner after the 3.3.0 release?

---

*If anyone ~~are~~ reading this is using Ledger's Python API with Python 2 I'd like to hear from you to learn more about your use-case and requirements. 🙂

[jwiegley]

I'm fine with dropping support for python2, but I agree with @tbm's comment that a deprecation notice and waiting at least one release cycle would be good.

[afh]

Glad to hear you are fine with dropping support for Python 2, @jwiegley. :)

It seems we are all in agreement that the next release should continue to include support for Python 2 along with a deprecation notice. And the release after the next will remove support for Python 2.

All that seems to be left to do is (proposing responsible people in parentheses):

• [ ] Write a deprecation notice (@afh, @tbm)
• [ ] Add and deploy deprecation notice to ledger-cli.org website and this repo (README, Release Notes, CHANGELOG, etc.) (@afh)
• [ ] Announce Python 2 deprecation plans on Ledger mailing-list (@tbm)
• [ ] Decide on the versioning for the subsequent releases (see my earlier comment) (@afh, @jwiegley, @tbm)
• [ ] Release a new version of ledger (@tbm)
• [ ] Merge this PR (@afh)

Do you agree with the above proposal? Did I miss anything?

[afh]

What are your thoughts on the following deprecation notice?

NOTA BENE: This is the last release that will support Python 2! With the next release ledger's' Python module will require Python 3.

The reason for removing support for Python 2 is that support for Python 2 ended on Januar 1, 2020 and the final release (2.7.18) published on April 20, 2020 marked the end-of-life for Python 2.

[jwiegley]

I'd really like to drop Python 2 support, so hopefully we can get the deprecation notice into the upcoming release, @tbm?

[tbm]

Honestly I'm ok with dropping Python 2 support now (sorry, I think I was the one wondering if we should wait another release; but it's 2023 now!).

But yeah if someone wants to add a deprecation notice.

Anyway, this PR no longer applies cleanly.

[afh]

I'm also okay with dropping Python 2 in the coming release and not wait for another release cycle. I haven't been following the mailing list or IRC too closely; has there been some conversation around this?

Once I'm done with rebasing this PR's branch onto master and resolving the resulting merge conflicts I'll push new changes to this PR.

[afh]

Odd the changes succeeded building and testing for me locally using nixpkgs; I'll investigate what is going (wr)on(g) with the checks here…

[afh]

My build succeeded locally despite the compiler error as my flake.nix still declared usePython = false 🤦

[afh]

ℹ️ The failing test on Linux 342 - RegressTest_786A3DD0 has been reported in #2176 and #2178 is an attempt to find the underlying cause.

[afh]

ℹ️ Failing test should be resolved once #2178 is merged.

[tbm]

This branch cannot be rebased due to conflicts

I've merged this manually. I hope I didn't mess up anything.

[afh]

Thanks for merging, @tbm 🙏

After all this time thinking I'd never see the day… Thank you, Ledger Python 2 support, for having served so well for so long 🥲