Hello,

Le jeu. 1 janv. 2026, 05:29, Dima Pasechnik <[email protected]> a écrit :

> On Wed, Dec 31, 2025 at 4:48 PM Grégory Vanuxem <[email protected]>
> wrote:
> >
> > Not related. GAP has a lot of work to be done. Particularly in the Julia
> GC world. It's a pity, they are not usable for example using libjulia.
> Inconsistent things I think. Oscar for example is not usable because of GAP
> through libjulia.
>
> Oscar doesn't really rely on GAP as a backend, as far as I know.
> Thus saying "Oscar is unusable because of GAP ..." doesn't make sent
> to me.


GAP is mandatory for Oscar  (on Julia).

Anyway,
> unless you run Oscar on a platform on which it's not really tested
> (Windows, of Oscar from GAP),
> it's hard to understand what you mean.
>

 I run it in WSL.  It is may be a problem.

Oscar works just fine on Linux and macOS, is well-supported and being
> worked on a lot.
> (while GAP is mostly in maintainance mode as far as its core
> functionalities are concerned)
> Few times I needed to run Oscar, it all went very well.
>

I discussed this about the memory management and future work, apparently,
on Discord. And this is a point for me. GAP plays apparently with the Julia
GC.

This is not the point here I agree. Take this as a after first year idea
please.

Greg

PS / Keep up the good work Dima you are doing an amazing work for sagemath.
Respect and, frankly speaking, have good things for this 2026 year.


> Dima
>
> >
> > Bad usage
> >
> > Le mer. 31 déc. 2025, 22:34, Marc Culler <[email protected]> a
> écrit :
> >>
> >> I am building an AppImage for Sage 10.8.  I have successfully built
> Sage 10.8 on my manylinux docker image and I can run it on my Ubuntu 24.04
> system.  But I have run into issues with libgap, and possibly other parts
> of sage.
> >>
> >> First, I cannot build the documentation because segfaults occur during
> the doc build.  Actually there are several, many of which occur during
> sphinxplot.  But I was able to track down at least one segfault in gap/
> element.cpython-313-x86_64-linux-gnu.so.  Based on disassembly, it seems
> to happen in make_gap_list.
> >>
> >> I then tried running some example code from libgap, and I get weird
> errors.  For example:
> >> libgap(vector((1/3, 0.8, 3)))
> >> produces:
> >> GAPError: Error, Variable: <<unknown>> must have an assigned value
> >>
> >> On my 10.7 AppImage, which was built on the same docker image, the
> result of that command is, as expected:
> >> [ 0.333333, 0.8, 3. ]
> >>
> >> For another example, the command:
> >> libgap.AlternatingGroup(4)
> >> produces
> >> GAPError: Error, no method found! Error, no 1st choice method found for
> `AlternatingGroupCons' on 2 arguments
> >>
> >> Again, that works fine on the 10.7 Appimage, producing the expected
> result:
> >> Alt( [ 1 .. 4 ] )
> >>
> >> And, to add one more, the command:
> >> sage: libgap.eval('5/3 + 7*E(3)').sage()
> >> produces:
> >> GAPError: Error, CoeffsCyc: function is not yet defined
> >> while the following command seems to work:
> >> libgap.eval('5/3 + 7*E(3)')
> >> 16/3*E(3)-5/3*E(3)^2
> >>
> >> Does anyone know anything which might give a hint about what is going
> on here?
> >>
> >> - Marc
> >>
> >> --
> >> 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 [email protected].
> >> To view this discussion visit
> https://groups.google.com/d/msgid/sage-devel/0de5b35b-2ac5-4987-97d5-b8cdfb2e301bn%40googlegroups.com
> .
> >
> > --
> > 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 [email protected].
> > To view this discussion visit
> https://groups.google.com/d/msgid/sage-devel/CAHnU2dbMmFAUZuHyXW1xEOKxbewEDtN3Uv53%3DjtrMyOif_UQsg%40mail.gmail.com
> .
>

-- 
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 [email protected].
To view this discussion visit 
https://groups.google.com/d/msgid/sage-devel/CAHnU2daWA%3D75RQrnn-z-74LtpNTK2mMPC3aWhs94NZeRBcKZvg%40mail.gmail.com.

Reply via email to