Maintainership of mpmath

[Fredrik Johansson]

Hi all,

I've had limited time to maintain mpmath recently (well, the last few years) due to a growing list of other projects, and it's clear that mpmath is suffering as a result.

I would be happy if someone else - ideally more than one person - could take over the maintenance responsibilities (reviewing and merging pull requests, making releases, and maintaining the documentation, tests and website).

I'd like to float the idea of transfering the mpmath project to the SymPy organization to be maintained by the SymPy developer community. In practical terms, this would mean:

* Transferring the mpmath repository on GitHub to the sympy GitHub org (I believe this is easy)
* Transferring ownership of pypi/mpmath
* Creating a repository for the mpmath.org website and hosting the website on the SymPy server (I could transfer ownership of the mpmath.org domain itself or just remain owner and point it to the new server)
* Transferring ownership of the mpmath mailing list

My second choice would be setting up a new GitHub org for mpmath and hopefully finding some volunteers to staff it. I prefer going with the existing SymPy organization (if willing) considering that the SymPy community is very active and that the projects are closely related. I'm hoping that the move could encourage contributors to make some long-overdue improvements to mpmath (performance, features, code clarity, ease of use) that will benefit both projects.

Note that this is not a proposal to merge mpmath back into the SymPy library as a submodule (which I'm firmly against).

Even if I step down as maintainer, I will continue to participate in discussions and code review (without claiming veto rights) and submitting patches when I have time.

Best,
Fredrik

[Aaron Meurer]

I'm in favor of seeing mpmath move under the SymPy organization. I'd
like to see more development work be done on mpmath. There are parts
of it that are underdeveloped. I also think that there's a lot of
numerical code in the parts of SymPy that wrap mpmath that would be
better if it lived in mpmath itself. There's also other instances
where SymPy does the wrong thing and mpmath does the right thing, for
example https://github.com/sympy/sympy/issues/19991.

Regarding the website, SymPy doesn't have a "server". We just use
GitHub pages. It works well for static websites because you don't have
to maintain any servers and it is free. The SymPy domains themselves
are managed by NumFOCUS.

I believe the actual transfership would be easy. The hard part is
figuring out if that's really what you want to do. Also, if mpmath
does start being more actively developed and new releases start
happening, we will need to decide how to manage that on the SymPy
side. Right now we don't have to worry about it because there's
basically only one mpmath version, but if mpmath were to release more
often, it opens up questions for SymPy like which versions we should
test against, if we should pin the mpmath versions, and if they should
be released in tandem or if independently is OK.

Aaron Meurer

> --
> You received this message because you are subscribed to the Google Groups "mpmath" group.
> To unsubscribe from this group and stop receiving emails from it, send an email to mpmath+un...@googlegroups.com.
> To view this discussion on the web visit https://groups.google.com/d/msgid/mpmath/CAJdUXTL9aDnECcivMvtXD0MpeLRPy1L9fTqADKegZKGD_QE2%2Bg%40mail.gmail.com.

[Rocky Bernstein]

While, I can't commit to anything long term, in the short term I would be happy to do the largely mechanical steps outlined in https://github.com/fredrik-johansson/mpmath/issues/584 to ensure that the current release status consistent between the homepage, github releases pages and README.rst. 

In doing that using the magic of of this wonderful feature called a hyperlink we can reduce inconsistency in future releases. 

If there really isn't much activity that occurs for whatever reason, even this one-time work would have a long term effect.