SCM_DEFER_INTS in 1.6

SCM_DEFER_INTS in 1.6

[Kevin Ryde]

What does SCM_DEFER_INTS actually do in 1.6?  I'm looking at the way
guile-gtk 1.2 puts it around gtk/gdk function calls, but I couldn't
spot anything in it which seemed relevant.

scm_ints_disabled gets set, but I couldn't see anywhere that variable
is tested, only take_signal() where it suppresses an SCM_NEWCELL (and
the result of that newcell seems unused anyway).

I'm wondering if the right thing would be just to drop the
SCM_DEFER_INTS from the guile-gtk generated code.  (The code there may
still want a mutex then, but that's a gtk/gdk matter, unrelated to
what guile might or might not need.)


_______________________________________________
Guile-devel mailing list
Guile-devel@gnu.org
http://lists.gnu.org/mailman/listinfo/guile-devel

Re: SCM_DEFER_INTS in 1.6

[Ludovic Courtès]

Hi,

Kevin Ryde <user42@zip.com.au> writes:

> I'm wondering if the right thing would be just to drop the
> SCM_DEFER_INTS from the guile-gtk generated code.  (The code there may
> still want a mutex then, but that's a gtk/gdk matter, unrelated to
> what guile might or might not need.)

I think what you see gets generated by G-Wrap.  Whatever
`SCM_DEFER_INTS' actually does, I can't see any reason why G-Wrap should
issue it as it does, and I've been tempted to remove it from there.  I
don't think it even needs to be replaced by a mutex or some such.  So I
think we should just happily remove it (from G-Wrap-generated code) and
see what happens!  ;-)

Thanks,
Ludovic.


_______________________________________________
Guile-devel mailing list
Guile-devel@gnu.org
http://lists.gnu.org/mailman/listinfo/guile-devel

Re: SCM_DEFER_INTS in 1.6

[Kevin Ryde]

ludovic.courtes@laas.fr (Ludovic Courtès) writes:
>
> I think what you see gets generated by G-Wrap.

Actually in guile-gtk 1.2 it's the old way with the build script
thingie.

> Whatever `SCM_DEFER_INTS' actually does, I can't see any reason why
> G-Wrap should issue it as it does, and I've been tempted to remove
> it from there.

Yep.

> I don't think it even needs to be replaced by a mutex or some such.

Well, gtk isn't reentrant so it needs some care.  I'm thinking to
leave it to the application.  But that's a subject for the guile-gtk
list.


_______________________________________________
Guile-devel mailing list
Guile-devel@gnu.org
http://lists.gnu.org/mailman/listinfo/guile-devel