xscanimage spins when scan cancelled

Graham Stoney (greyham@research.canon.com.au)
Thu, 15 Apr 1999 13:36:49 +1000 (EST)

G'day SANE developers,

Firstly, congratulations on the fine work you folk have been doing as part of
the SANE effort. It's certainly an impressive work, and a nice clean interface.
May you all live long and prosper.

Unfortunately I have a problem I was wondering if anyone could shed some light
on. xscanimage hasn't been my friend today, and I'd like to make amends. The
problem I'm having is that cancelling a scan doesn't work. My sane_cancel
function gets called and xscanimage stops calling sane_read, but the "Scanning"
panel with the progress bar never disappears. Attaching to xscanimage in
gdb reveals that it's spinning around g_main_iterate chewing CPU time.
Continuing to press the "Cancel" button on the "Scanning" panel causes
sane_cancel to be called again, but my backend has well and truly stopped by
this time, so it has no further effect.

I tried upgrading libg and libgtk to 1.2.1, but with no improvement. Short of
continuing to tear my remaining hair out or embarking on a multi-year course of
study in the finer points of gtk, does anyone have any suggestions as to what
might be causing this? I'm on a Red Hat 5.1 Linux system with sane 1.0.1-pre4.

Thank you in advance,
Graham

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