Skip to content
Pierre Letouzey edited this page Oct 18, 2017 · 113 revisions

This page is used to organize the next Coq Working Group (in French GT Coq). The framadate link to decide which day it will happen was:

https://framadate.org/2a2xL73ZXlaYpNWK

Organization

The next Coq Working Group will take place on February 16th at Sophie Germain, PPS on the 3rd floor.

Schedule

  • 10:00 - 11:00 Development rules and best practices (Maxime)
  • 11:00 - 12:00 Discussion on tactics, ssreflect (Hugo)
  • 12:00 - 12:10 Syntax extension : '(pattern), e.g. "Definition f '(x,y) := ..." - (Daniel)
  • 12:10 - 12:20 Splitting libraries (Emilio)
  • 12:20 - 12:30 Bug management systems (Emilio)
  • 12:30 Lunch
  • Afternoon: Pull requests

Some more possible topics (by Emilio J. Gallego)

Bugzilla vs other tools?

I'm not an expert bugzilla user, but I find for instance GHC's setup a bit more welcoming:

In particular I find it easier to:

  • get an overview of the bugs that need most help, that are untriaged, obsolete, etc...
  • understand the roadmap, see what's going to be merged in the next release. (So I can adapt/test against)
  • merge directly from bugs/put link to actual code/commits, etc...
  • comment on commits/branches.
  • do code reviews for patches.
  • personalize the interface.
  • see what others are doing/working on.

I wonder which items can be done with bugzilla setup improvements and which ones are inherently outside the capabilities of the tool.

Regarding github's/gitlab, IMHO their interface currently lacks enough release management tools, so I'm not much of fan (other than for convenience).

Another interesting approach is to have every pull request to go by a mailing list review as done in linux.

README.dev

In particular it would be nice to have a small "How to contribute to Coq, guide for the casual developer" document.

Splitting the libraries

As discussed this summer, would a separate release process for the libraries and for the compiler make sense?

IMHO, this would imply a lot of effort, but on the other hand it could bring significant advantages.

GM Personally, I think this would be a very good thing. Three reasons: 1) it would likely improve the adoption of opam. 2) it would allow additions to the library without needing to get them merged into Coq. 3), I think it would allow help getting more people involved without Coq development without them needing to work with the Coq internals (yes you can submit pull requests to the standard library without touching the Coq implementation but it seems like a lot). If this gets done, there should be something like a 'batteries' package that packages up some of the most commonly used pieces of the library, in particular, this is probably useful for people where opam does not work well.

GM One way to see if this would be beneficial would be to poll the community and see how much interest there would be from others in working on this. If no one would want to contribute then it probably is not worth it.

coqdoc future

What are the plans for CoqDoc_? What about supporting writing proof scripts in rich markup languages?

Minor points

Coq mailing lists are not indexed by google!

Should cleanups target master or 8.5?

Why do we have "trunk" instead of "master"? Technically, it does not matter. It just sounds (to me) curiously in the git-context.

Clone this wiki locally