Point users to PythonCall.jl

[MilesCranmer]

@mkitti What do you think? Currently there is nothing pointing users to PythonCall.jl besides comments in GitHub issues. A new user might see this repository having more stars => preferred/stable option, despite PythonCall.jl being the one that is actively maintained.

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Comparison is base (8ab68f4) 87.52% compared to head (27dea0c) 79.60%.

Additional details and impacted files

@@            Coverage Diff             @@
##           master     #556      +/-   ##
==========================================
- Coverage   87.52%   79.60%   -7.92%     
==========================================
  Files          39       39              
  Lines        2349     2349              
==========================================
- Hits         2056     1870     -186     
- Misses        293      479     +186     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[mkitti]

I think we should involve a lot more people before proceeding with this.

[MilesCranmer]

Fair enough but this has already happened in the various GH issues (eg #555 (comment), #542 (comment)).

I don’t see a reason it shouldn’t also be mentioned on the README so users don’t accidentally use unmaintained software. Why should they need to post an issue before we tell them to use PythonCall?

[jlapeyre]

It seems clear to me that a link the PythonCall would be useful.

You could debate about what should be said about the two projects. But just about anything is better than nothing.
The text of this PR looks ok to me.

[MilesCranmer]

Seems people are happy with this in https://discourse.julialang.org/t/deprecate-pyjulia-in-favor-of-py-juliacall/111634. Are we good to merge @mkitti? I wouldn't consider this an 'official deprecation' but just a helpful pointer. Maybe I could say

- interface has transitioned to PythonCall.jl
+ interface has mostly transitioned to PythonCall.jl

to be even more explicit that this is not an official deprecation, if you would like?

[jlapeyre]

Just to reiterate and agree with @MilesCranmer . The main points should be easy and not contentious. Add text to landing pages, particularly those with high search visibility that give links to related projects with a bit of context. And this PR hits the main point.

The question of in some sense deprecating pyjulia and anything else that requires any amount of real work can easily be postponed.

Ideally all of the julia/python interface repos should have a section similar to this one