[AG-TECH] New Fedora Core 2 AGTk CVS snapshot

[Douglas Kosovic]

"Eric Olson" <eolson at mcs.anl.gov> wrote:
> Hi Fred and Doug,
> I checked in a small change to VideoService.py and VideoProducerService.py
> in cvs that should fix the problem causing the video port settings getting
> reset to the default port all the time (Bug #1067).
> Let me know if it fixes that problem for you once you're able to test
> it.

I'm fairly confident I've fixed the VideoService and VideoProducerService
vic default port problem now. New versions of VideoService.py and
VideoProducerService.py can be found in :

  http://www.itee.uq.edu.au/~agn/ag_fedora/FC2/unstable/patches/

They are complete files rather than patches. The main problem with the patch
I posted last friday was that I set the input port after vic had started
transmitting, also I didn't check if the GUI had finished initializing.

Bug #1067 is due to vic not identifying video devices by their device name
(e.g. "/dev/video2/"), but by a nickname, an example of which looks like:
  "v4l- BT878(Pinnacle PCTV Studio/Ra) /dev/video2"

Which means the following line doesn't set the default port to S-Video for
dev/video2:
   set defaultPort("/dev/video2/") S-Video
but the following does:
   set defaultPort("v4l- BT878(Pinnacle PCTV Studio/Ra) /dev/video2")
S-Video

My solution involves setting the input port in the user_hook procedure (of
the vic startup file) where the input device has already been assigned, so
there is no need to specify the device when setting the input port.

Also note on Linux,  VideoService and VideoProducerService use the
VIC_DEVICE environment variable to tell vic which device to use, I've added
code to check if the VIC_DEVICE env var isn't set (like when vic is invoked
on the command-line), then a substring match search between the device name
and the nicknames is performed to determine which device to use.


Cheers,
Doug