<!-- Some styling for better description lists --><style type='text/css'>dt { font-weight: bold;float: left;display:inline;margin-right: 1em} dd { display:block; margin-left: 2em}</style>

   johanhelsing: <u>mchehab</u>: how could the compilation of libmediactl affect that hack? It's only using the header file of mediactl to get the declaration to decide which overload to call, so as far as I understand, so it will always be the c++ compiler that compiles Qt that decides which version to use.
   ***: prabhakarlad has left
   mchehab: <u>johanhelsing</u>: ah, I see
   <br> ndufresn`: did a quick look at --enable-v4l2-probe code at v4l2 1.4.4 plugin...
   <br> I can't see how it would get any delay from *any* V4L2 camera
   <br> if the device is not a M2M, the only thing it does it to open the device and do a VIDIOC_QUERYCAP
   <br> what camera is having a long probe time? Is it an OOT driver?
   <br> ndufresn`: btw, adding bayer format to vim2m makes v4l2 probe to not expose it as a videoconvert source
   <br> ndufresn`: it seems that the problem is that it assumes that any m2m transform device can do a any to any conversion:
   <br> gboolean
   <br> gst_v4l2_is_transform (GstCaps * sink_caps, GstCaps * src_caps)
   <br> {
   <br> gboolean ret = FALSE;
   <br> if (gst_caps_is_subset (sink_caps, gst_v4l2_object_get_raw_caps ())
   <br> &amp;&amp; gst_caps_is_subset (src_caps, gst_v4l2_object_get_raw_caps ()))
   <br> ret = TRUE;
   <br> return ret;
   <br> }
   <br> the logic should be instead, if !encoder &amp;&amp; !decoder, then it is a transform
   ***: benjiG has left
   <br> prabhakarlad has left