Gerd Knorr wrote:
BTW: Someone at the ieee1394-devel ml wrote, that a main reason to
create a new interface (-> the new dv1394 module) was that V4L2 does not
report dropped frames during capturing. Is this true? If so, that might
be something to fix before merging into the kernel...?
That was true for the old v4l interface. v4l2 has timestamps and
sequence numbers for video frame buffers.
Thank's for the reply. So there should be nothing which prevents V4L2 from
proving the complete dv1394.o functionality. (They have a slightly different way
to manage the transfer buffers, but that' not critical , I think).
For other comments on the API look check the recent discussions on the
v4l2 api last days. For DV data we likely have to add V4L2_PIX_FMT_DV.
You might also have a look at the current "struct v4l2_timecode", DV
likely is the first one who might use it. If there is something wrong
we should fix it before submitting v4l2 into 2.5.x ...
I don't think that I'm the right one (yet) to find bugs in there, sorry. If I'll
really be the one to write an adapter between dv1394.o and videodevX.o, I most
likely won't start with the programming until next year ...
Regards,
Norbert
