Coercion to interfaces

[Travis Scrimshaw]

Hey all,
   On https://trac.sagemath.org/ticket/18266, Vincent and I are having a discussion about how much coercion an interface, GAP in this particular instance, should have. PARI currently has everything being able to coerce in. GAP currently only has select rings, but Vincent's proposal is to mimic PARI. Vincent's argument is that this requires less maintenance, whereas I believe it is far too generous and results in an incorrect error message (but it does error out in both cases the examples I tested). We are hoping you could weigh in with your thoughts.

Best,
Travis

[Dima Pasechnik]

I believe that the work on libGAP interface might better wait for GAP folks to come up with their own libGAP. They refused to merge "our" libGAP, and they say they have their own version in the works, which they need to create a Jupyter kernel for GAP.

Dima

[Vincent Delecroix]

I don't understand what this point has to do with the OP (which is a
clear question about *coercion* and *interfaces in general*). libGAP was
just an *example*.
However, whatever libGAP will be it will remain a C library. How it is
integrated in Sage and how it interacts with the rest of Sage is a Sage
question...

[William Stein]

On Sun, Jul 24, 2016 at 6:38 AM, Vincent Delecroix

Sage's libGAP is 7 years old, and has been heavily tested, rewritten
multiple times (me, Robert Miller, Volker Braun), and is likely much
more mature and robust than something brand new.

I don't know about the politics of 'They refused to merge "our"
libGAP' (though I'm curious). When I wrote the first libGAP it was in
Barcelona in June 2009, and the Singular talk at the same conference
(MEGA) was mostly about how excited the Singular devs were about the
new C++ library interface (a sort of libSingular) that we (mainly
Martin Albrecht) had contributed to Singular. I started libGAP a
little bit inspired by that.

-- William

>
> --
> 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 post to this group, send email to sage-...@googlegroups.com.
> Visit this group at https://groups.google.com/group/sage-devel.
> For more options, visit https://groups.google.com/d/optout.



--
William (http://wstein.org)

[Dima Pasechnik]

Upon reading of the trac ticket I got an impression that while an extension you propose

is very desirable, it would need more work on the C side of libGAP, as it does not expose

enough of GAP's C functions.

Dima