> Negative. It is buggy as hell. I am fixing this, but I have almost no time
> to do it. Someone added really hard-to-implement options (like the
> geometry stuff) there, and then didn't do it right.
> Of course I get the bad credits for that, as I'm listed as the author.
Cool. I just wanted to make sure I wasn't going insane :-) I'm considering
writing a "reference" backend that doesn't actually rely on PNM or the
like, but generates datastreams so you can debug that you are, indeed,
getting what the backend is giving :-)
In particular, I'm trying to diagnose an oddity in my current code. I'm
using the umax driver, and scanning with mode='Lineart'. The data comes
back to me in GRAY frames, with depth=1. This is fine. So I take the data
as-is, and dump it to a RAWPBM format PNM file. The image is wrapped by ~
1 byte horizontally. Oddly, running the same scan with scanimage produces
a "correct" image.
Also oddly, my unpacking for 8-bit GRAY and 8-bit RGB frames seems to
generate correct images.
So, you can see that having a very controllable datasource would be
helpful for debugging this sort of thing :-) I could put in a predefined
bitstream and know /precisely/ what the driver was going to give me :-)
(Okay, okay, I know you're saying that I could get the same results with a
well-known PNM and a driver without 'helpful additions' :-))
Anyway, back to hacking.
-- Tripp Lilley + Innovative Workflow Engineering, Inc. + (tripp@iweinc.com) ------------------------------------------------------------------------------ "So perl is like a graceful swan, gliding across a lake. But underneath the water, there's these little legs, paddling like buggery."-- beable van polasm in http://www.dejanews.com/getdoc.xp?AN=440381254
-- Source code, list archive, and docs: http://www.mostang.com/sane/ To unsubscribe: echo unsubscribe sane-devel | mail majordomo@mostang.com