Stefan Richter wrote:

Perhaps you have two different versions of libdc1394 installed, and
coriander is being linked against the old one.  Or this libdc1394
instance was configured with the juju backend at build time, which would
be kind of sad.  Or you are running coriander as a user who does not
have permission to access /dev/fw1 (or whichever file represents the
camera at the time --- fw1 would be typical though).

SVS doesn't work with the new stack.

Any idea why not?  If it is libraw1394 based, it should just continue to
work (in theory) independently of which kernel drivers are present, if
SVS was linked against libraw1394 v2 (ideally the newest one).

One version of libdc1394 is installed. I built it from source.

I had set /dev/fw1access so everyone has rw access.

SVS is linked against the old libraw, and the source for the libraries is not provided, unfortunately.

When I try to rebuild coriander with the new kernel, I run into package inconsistencies. Maybe I'll just have to wait a few more versions of linux until things are more stable.

David Moore wrote:

On Mon, Jun 14, 2010 at 1:44 PM, Paul Benjamin <dpaulbenjamin@gmail.com>wrote:

My problem now appears to be in userspace. I cannot get coriander to see
the new
stack (it always asks for ieee1394). I must not be building it properly.
SVS doesn't work
with the new stack. I'm not sure how to get video from the camera with the
new stack.


If you set the environment variable DC1394_DEBUG=1 prior to running
coriander, you should get some extra debugging information generated from
libdc1394.  This should help narrow things down.

Did not have any effect for me.

Also, consider using the example programs such as "grab_gray_image" that
come with libdc1394.  That way you can figure out if libdc1394 or coriander
is the problem.  By the way, with the new stack, libdc1394 uses the kernel
directly and does not go through libraw1394.

grab_gray_image gives:
Using camera with GUID 1687300011660
libdc1394 error: Could not get a valid M0N08 code

Thanks,
Paul