[jitsi-dev] [PATCH] Linux local video during a call


#1

Hi all,

As some of you may know, in a video call on Linux we have only the video of the remote peer but not the local ones. Users of Windows and Mac OS X version of Jitsi do not have this problem.

Finally I find some time to work on this issue. To summarize the problem is located in the Cloneable* stuff inside FMJ. It will only create Push* like streams (even if we pass a PullDataSource). I modify FMJ in order to correctly create a valid Pull-like clone (with associated PullBufferStream slaves) for our Video4Linux2 PullBufferDataSource capture device. As I have no way to test PullDataSource devices with cloneable stuff, I have only modify the behavior for PullBufferDataSource capture device.

In attachment, you will find one patch for FMJ and one for Jitsi (which basically bring back the possibility in VideoMediaDeviceSession to create a clone capture from Pull* like capture device).

Lyubomir, can you please review the patch when you have time, and tell me what do you think about the FMJ modifications ?

Best regards,

fmj-cloneable-pullbufferdatasource.diff (6.52 KB)

jitsi-cloneable-pullbufferdatasource.diff (904 Bytes)

···

--
Seb


#2

Hey Seb,

Thank you very much for the contribution! I'll gladly review the patches, I just have to wrap up my current development related to the video conferences.

Best regards,
Lyubomir

···

On 15.03.2012, at 19:41, Sebastien Vincent wrote:

Lyubomir, can you please review the patch when you have time, and tell me what do you think about the FMJ modifications ?


#3

Hi Seb,

Thank you very much for the contribution! I've committed your patches
in Jitsi r9499. With respect to the FMJ patch, I hope you will not
mind that I took the liberty to simplify it (because FMJ was handling
the Pull*DataSource/Stream cases already), I tried to fix the
non-Buffer cases and I removed an unused method or two.

Best regards,
Lyubomir


#4

Hi Lyubomir,

Great, thanks :).

Best regards,

···

--
Seb

Le 03/04/12 10:13, Lyubomir Marinov a �crit :

Hi Seb,

Thank you very much for the contribution! I've committed your patches
in Jitsi r9499. With respect to the FMJ patch, I hope you will not
mind that I took the liberty to simplify it (because FMJ was handling
the Pull*DataSource/Stream cases already), I tried to fix the
non-Buffer cases and I removed an unused method or two.

Best regards,
Lyubomir