[sane-devel] First version of the new test backend released

[Henning Meier-Geinitz]

Hi,

On Sun, Mar 31, 2002 at 10:21:06PM +0200, mh wrote:
> One remark:
> You allow a three-pass/handscanner simulation in color mode with a bit depth 
> of 1 (which crashes xsane and xscanimage and lets quiteinsane loop forever). 

I can't reproduce this with xsane 0.84 in viewer mode. The scanned
image is wrong (if it's not completely black) but it doesn't crash
even if I enable both three-pass and hand-scanner simulation.
When scanning the preview it complains that it can't handle bit 
depth 8 :-)

xscanimage (CVS) refuses to preview without any options set but shows
"something" in three-pass mode. When scanning it "works" with and
without three-pass mode, however, the pnm images aren't correct.

With quiteinsane 0.4.1 I get an error message without any error text
when trying to scan in 1-bit color mode even if I don't select
three-pass and/or hand-scanner simulation.

Do you use any other special options?

> Is this just a bug in the test backend or one of the tests ? (Most probably, 
> of course, it's an attempt to drive frontend authors insane :-)

1-bit RGB *is* insane. However, the SANE standard allows this
combination, at least that's my conclusion. The mustek backend
supported this mode years ago, because the scanners can do it natively.
I removed it because I thought it doesn't make much sense.
So I guess the best thing is to just refuse to scan in 1-bit RGB.

hand-scanner and three-pass together probably won't be found in the
wild. But again, I think it's allowed by the SANE standard as long as
the frames stay the same size.

There may be a bug in the test backend especially in the 1-bit RGB
mode. I can't really test this mode because no frondend seems to
really support it :-)

Can you try to locate the position in the test backend where it loops
or breaks (SANE_DEBUG_TEST=255)?

Bye,
  Henning