Hi Steve
I have a couple of other ideas on the RGBRouter problem.... as I understand
it, the router was designed to perform the data unpacking from scanner
buffer format to the internal RGB sane buffer format while at the same time
accommodating any color offset shifts needed (spatially aligning RGB). The
new code uses offsets reported by the scanner... It could be possible that
the snapscan 1212u is reporting bogus offsets that are getting the router
totally confused... I will try and figure out how to ignore the offsets from
the scanner and see if that makes things better. If you could describe how
the elements of the rgbrouter structure are supposed to be used, it would
help me decipher what is going on.
An update on my earlier comment... the scanning of small color image at
50dpi
doesn't really work.... it doesn't hang xscanimage but the generated file is
missing about 10-20% of its bytes and fails to open in gimp with an
premature EOF message.
I feel this is directly related to the rgbrouter problems. I'm getting quite
an education in the sane code... its not one of the easier pieces of code to
get ones mind wrapped around with all the nested virtual functions.
Thankyou for your comments... I'll post some dumps of the router related
stuff later today.
=======================================================
Lawrence Glaister VE7IT email: lg@jfm.bc.ca
1462 Madrona Drive http://jfm.bc.ca
Nanoose Bay BC Canada
V9P 9C9
=======================================================
----- Original Message -----
From: Steve Underwood <steveu@coppice.org>
To: <sane-devel@mostang.com>
Sent: Thursday, September 21, 2000 7:48 AM
Subject: Re: problems with 1.0.3 and color scanning
> Lawrence Glaister wrote:
>
> > I am trying to get sane 1.0.3 working with my snapscan 1212u usb
scanner.
> > The scanner works properly under 1.0.1 but fails in various ways under
1.0.2
> > and 1.0.3 and the cvs code as of sept 19. I have narrowed the problem
down
> > to the snapscan-sources.c RGBRouter_remaining() function.
> > - grayscale scanning works fine
> > - color scanning a small image at 50dpi works
> > - color scanning same image at 75dpi breaks xscanimage
> > by broken, I mean that xscanimage enters into an infinite loop using
100%
> > cpu time.
> > the cause for this appears to be a negative result returned by
> > RGBRouter_remaining(). There are two possible return values calulated in
> > RGBRouter_remaining() and one of them always looks like the correct
> > answer... the problem comes in that the conditional used doesn't seem to
> > work in all cases (sometimes the wrong value is returned and when this
> > is -ve, all hell breaks loose.). I have been getting a headache trying
to
> > decipher how the new RGBRouter code is supposed to work... could the
author
> > (pere) or someone have a look at the function and see if they see any
> > obvious errors. I have not seen a lot of postings about 1.0.3, so the
> > problem may be related only to the snapscan backend with a usb patch
applied
> > to it.
> > I can supply some rather detailed dumps of the good scans and bad scans
as I
> > have added a lot of dbg calls into the snapscan code.
> >
> > Are there any snapscan scanners usb or scsi out there working with sane
> > 1.0.3???
>
> Excellent. This look likes a really solid lead for what goes wrong. I will
try
> to look at the RGBRouter code in the next couple of days, and see if I can
spot
> what goes amiss. Actually, I thought the current RGBRouter code was fairly
> straightforward, but maybe that's only to my own twisted mind! The
existing
> code certainly does work for some scanners. It works just fine for my Acer
> 610+, and a few other people have reported success with some scanners.
>
> Lets hope this is the root cause of several not quite identical problems
people
> have reported.
>
> If you have any debug logs, or other useful info please mail it to me. my
code
> hasn't given me any trouble so far, so a log of an actual failure would be
> helpful.
>
> Steve
>
>
>
> --
> Source code, list archive, and docs: http://www.mostang.com/sane/
> To unsubscribe: echo unsubscribe sane-devel | mail majordomo@mostang.com
>
-- Source code, list archive, and docs: http://www.mostang.com/sane/ To unsubscribe: echo unsubscribe sane-devel | mail majordomo@mostang.com
This archive was generated by hypermail 2b29 : Thu Sep 21 2000 - 08:11:24 PDT