Demote brial (= polybori) from standard to experimental

[Matthias Koeppe]

I propose to demote this package to experimental.

- It has been declared dead at least once -  https://martinralbrecht.wordpress.com/2015/06/13/polybori-is-dead-it-needs-your-help/

- It has no upstream maintainer (except for ermergency fixes by Francois Bissey) - https://github.com/BRiAl/BRiAl/graphs/contributors, https://sourceforge.net/projects/polybori/

- It has been dropped from Debian testing, where it seems to block SageMath upgrades (Sage is stuck at 9.5 in Debian/Ubuntu)

- The conda and homebrew packages of brial lead to segfaults (https://github.com/sagemath/sage/issues/35595, https://github.com/sagemath/sage/issues/34780)

- It is disconnected from the advances in SMT (satisfiability modulo theories) over the past decade (representative paper: https://arxiv.org/abs/2305.00028v2)

[François Bissey]

Please remove that pain from my a**.
Seriously, I am not sure why it leads to so many issues. I have been
combing the code on a regular basis for compiler warnings and stuff and
fixing them.
Debian has issue with the dead brial python interface package, which is
definitely unmaintained because we moved the essential bits inside sage.
Why they persisted with it, I do not know.
If people have issues, I welcome them opening them on the tracker so I
can look at them.
Nevertheless moving to newer, better maintained stuff would make my day.

François Bissey

> --
> You received this message because you are subscribed to the Google
> Groups "sage-devel" group.
> To unsubscribe from this group and stop receiving emails from it, send
> an email to sage-devel+...@googlegroups.com
> <mailto:sage-devel+...@googlegroups.com>.
> To view this discussion on the web visit
> https://groups.google.com/d/msgid/sage-devel/76bc224b-8ed0-4274-9811-1d118737a8edn%40googlegroups.com <https://groups.google.com/d/msgid/sage-devel/76bc224b-8ed0-4274-9811-1d118737a8edn%40googlegroups.com?utm_medium=email&utm_source=footer>.

[Dima Pasechnik]

good idea

--
You received this message because you are subscribed to the Google Groups "sage-devel" group.

To unsubscribe from this group and stop receiving emails from it, send an email to sage-devel+...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/sage-devel/76bc224b-8ed0-4274-9811-1d118737a8edn%40googlegroups.com.

[Matthias Koeppe]

A milder change, demoting it to optional, is ready for review in https://github.com/sagemath/sage/pull/36380

[Nathan Dunfield]

This makes sense to me.

Nathan

[Vincent Delecroix]

+1 for me too

> --
> You received this message because you are subscribed to the Google Groups "sage-devel" group.
> To unsubscribe from this group and stop receiving emails from it, send an email to sage-devel+...@googlegroups.com.

> To view this discussion on the web visit https://groups.google.com/d/msgid/sage-devel/427f1f70-7cc7-4f5e-8466-ad28bb81ae87n%40googlegroups.com.

[Matthias Koeppe]

I'll remark that the approach of my PR https://github.com/sagemath/sage/pull/36380 takes brial to the same level as the existing optional packages bliss, coxeter3, sirocco, tdlib, etc. and applies the same approach to the Sage library code depending on the library: It defines a separately buildable, pip-installable distribution sagemath-brial.

The rest of the Sage library, still shipped as sagemath-standard, is isolated from the problematic dependency on brial.

That's an illustration of how the modularization project improves portability and makes it easier for downstream distribution to carry the Sage library.

[Dima Pasechnik]

+1

[Matthias Koeppe]

Still waiting for review: https://github.com/sagemath/sage/pull/36380

I had hoped to get this into the Sage 10.4 release.

[Dima Pasechnik]

1) Brial is too broken for being an optional package.

2) It should be a separate PR, not one doing 10 other unrelated changes, and touching 600+ files in  the process