Re: umax-backend

Mark Cornick (msc@zorak.gsfc.nasa.gov)
Thu, 16 Oct 1997 17:21:09 -0400

On Thu, Oct 16, 1997 at 10:42:53PM +0200, Peter Daum wrote:
> Furthermore, I still keep getting segfaults right when the
> program starts (see the message I sent on 9 Oct 1997
> (Pine.LNX.3.96.971009143034.8415A-100000@swamp.bayern.net). Am I
> the only one having these problems?
>
> The crash itself always occurs in gtk_real_widget_queue_draw
> (gtkwidget.c: line 2813); immediate cause is an element with an
> invalid data pointer int gtkmain's "idle_functions" list.
> Although the error is pretty deterministic and reproducible, I am
> not able to find out, where exactly that corrupt address gets in
> there ...

I got behavior like this from xscanimage when I linked it against
gtk+970925 (before David sent the recent patch - I haven't patched the
gtk yet.) But:

1) It only bombs when xscanimage is started in the background
(i.e. "xscanimage &" from the prompt, or selecting from the gimp
menus.) If I do "xscanimage" from the prompt and keep it in the
foreground, I still get the message about gtk_real_widget_queue_draw,
but the program runs fine.

2) I didn't get this behavior when I linked it against an older gtk
(from gimp 0.99.10, w/ David's patches)

3) This is on my system at home, which has a Umax Astra 600S. On my
system here, which has a Color QuickCam but no Umax scanners, it works
every time.

--mark

-- 
Mark Cornick
LTPCF Workstation Support, Code 922
Hughes STX Corp / NASA Goddard Space Flight Center
mark.cornick@gsfc.nasa.gov



--
Source code, list archive, and docs: http://www.mostang.com/sane/
To unsubscribe: echo unsubscribe sane-devel | mail majordomo@mostang.com