* Marc Lehmann ([EMAIL PROTECTED]) [010114 09:21]:
> On Thu, Jan 11, 2001 at 10:22:04PM -0600, Rick Bradley <[EMAIL PROTECTED]> wrote:
> > I've run into a problem with gauss_rle and perl-called script-fu's in
> > Gimp 1.1.29.
> You might try to upgrade. The communication with script-fu from other
> plug-ins was broken until very, very recently, and I am not sure it works
> correctly now. It is unlikely to be a problem with perl as it proved to
> never be the problem so far (and it works with everything else).

I'll do that shortly (upgrade) and see what I get.  It's also been my
experience that, other than a deeply nested debugger bug I found
(which was immediately fixed) back in '96, Perl is generally Not The

> >  - On those script-fu's which internally use gauss_rle I get the
> > following error in the server output log:
> Try to call Gimp::set_trace(Gimp::TRACE_TYPE) before the call to script-fu
> and check the arguments. If these are correct then you might fail for the
> corruption bug in libgimp or something new.

I've got that in.  The args passed in seem correct, and for other
(similar) fu's things work well.  I'll see if the upgrade does the trick.

> (It could also be a bug directly in the scheme interpreter, but I think
> that's unlikely)

Yeah, that seems fishy to me too -- something fundamental enough to
cause this sort of problem would presumably break umpteen other

> However, it might also be a bug in the script-fu script. Sicne there is
> no form of type-checking it is very customary for script-fu scripts to
> only work directly after starting the gimp, or when certain conditions are
> met. For example, it could pass in a layer id instead of an image id.

Hm.  Will dig over them with a finer-toothed comb.  The same
script-fu's do work perfectly interactively so it's still presumably a
perl-fu<->script-fu interaction problem, and my #1 tactic is going to
be to do the gimp upgrade to see if that fixes the hoodoo.

  Rick Bradley                  http://www.roundeye.net

Reply via email to