[Pxc] I had the same trouble as Tye Brady. Any solution?

[Alessandro Rubini]

> First let me thank Alessandro for writing the device driver and the prompt
> response to my question.  It really helped me out (BTW his book is awesome
> on writing device drivers).

I'm not a native speaker, I can't understand if this is ironic or thankful :)

I don't think I managed to fix your problem, from my traces it looks like
I asked you to take some steps and send me result, and it all ended there.
 
> OK the solution for me was to pitch the "prebuilt" kernel for me that
> came with RH7.0.  Whatever was happening, it was incompatible [...]

Well, this is really good information. While I always suggest to take
a clean Linus kernel, I didn't know RH7.0 was so different in this
respect.  I fear I must try it out.

> This sounded easy enough but for a newbie like myself, I tripped multiple
> times on the job.

You can refer to http://www.linux.it/kerneldocs/kconf/ (ok, that's not
widely advertised, but I don't even know if it is really useful or not :).

Anyway, I'll put a big warning on the new version of the driver, due
out 2.5 weeks from now (this time I'm under contract by ImageNation to
add some features, so I'm bound to meet the deadline.

> I also use bttv0.7.59 with i2c-2.5.5 to run XAWTV and bttvgrab quite nicely.
> I believe they incorporated Alessandros drivers and work.

They only grabbed initialization code. Bttv is a completely different
code-base, it was not stable or commented enough to be a viable
approach when I had the need to support PXC200 in my software.

Now I'm able to run video4linux too (after a *huge* amount of hours to
make it work), so the next version of my driver will include
video4linux support, and I'll possibly contribute code to support the
PXC200-F in bttv (if I have time).

> My personal application is a machine vision like task of trying to measure
> the drift of a star in a CCD camera.  I would like to locate the object
> in the camera, stay locked on to it, and then measure how much it moves
> in a certain amount of time.  If anybody has any references to any code
> or solutions, that would be GREAT!

My "bisce" tracking program does almost the same. It is quite
"primitive" in design, but it works satisfactorily for the physiology
lab that uses it (I even send two computers abroad with "bisce"
preinstalled). Its docuemntation is at
http://gnu.systemy.it/software/bisce/ , download info at 
http://gnu.systemy.it/software#bisce

/alessandro