GL: tex_id is always zero after updating from 1.8 to 1.10

[Gottfried Haider]

Our code [1] is modeled after the testegl example in gst-omx. In 1.8,
in handling the "handoff" signal, we would retrieve a (nonzero) tex_id
from the GstBuffer we're handed.

With 1.10 the code doesn't throw any particular error, the callback is
still begin periodically executed - but the tex_id is always zero.

Any idea about what could have caused this change in behavior?

I am testing on macOS 10.11.6 with the gstreamer-1.0-1.10.0-x86_64
binary from the site.

Thanks in advance,
Gottfried
_______________________________________________
gstreamer-devel mailing list
[hidden email]
https://lists.freedesktop.org/mailman/listinfo/gstreamer-devel

[Nicolas Dufresne-4]

Le samedi 12 novembre 2016 à 01:32 +0100, Gottfried Haider a écrit :
> Our code [1] is modeled after the testegl example in gst-omx. In 1.8,
> in handling the "handoff" signal, we would retrieve a (nonzero) tex_id
> from the GstBuffer we're handed.

Those are far from being the best examples.

>
> With 1.10 the code doesn't throw any particular error, the callback is
> still begin periodically executed - but the tex_id is always zero.
>
> Any idea about what could have caused this change in behavior?
>
> I am testing on macOS 10.11.6 with the gstreamer-1.0-1.10.0-x86_64
> binary from the site.

Did you forgot to add glupload in your pipeline ?

Nicolas
_______________________________________________
gstreamer-devel mailing list
[hidden email]
https://lists.freedesktop.org/mailman/listinfo/gstreamer-devel

[Matthew Waters]

On 12/11/16 11:32, Gottfried Haider wrote:

> Our code [1] is modeled after the testegl example in gst-omx. In 1.8,
> in handling the "handoff" signal, we would retrieve a (nonzero) tex_id
> from the GstBuffer we're handed.
>
> With 1.10 the code doesn't throw any particular error, the callback is
> still begin periodically executed - but the tex_id is always zero.
>
> Any idea about what could have caused this change in behavior?
>
> I am testing on macOS 10.11.6 with the gstreamer-1.0-1.10.0-x86_64
> binary from the site.

Nothing comes to mind with this little information.

A small reproducible test case would be great to attach to a bug :)

Cheers
-Matt

> Thanks in advance,
> Gottfried




_______________________________________________
gstreamer-devel mailing list
[hidden email]
https://lists.freedesktop.org/mailman/listinfo/gstreamer-devel

[Gottfried Haider]

Hi Matt,

This is again the Processing (Java) library we're working on. So far I have only tested it on macOS with the binary distribution of the library, not sure if I'll find time to compile GStreamer 1.10 on Linux anytime soon.

Would a reproducer for macOS with Processing, and a couple of fprintf()s thrown in be sufficient?

Anything else I can do on my side? (output of a certain debug level with both 1.8 and 1.10?) I wish I could make this more self-contained, but I've never done GL in C... If it's worth testing, I could compile 1.10 for Linux on my laptop, and check if it happens there too.

Thanks

Gottfried

_______________________________________________
gstreamer-devel mailing list
[hidden email]
https://lists.freedesktop.org/mailman/listinfo/gstreamer-devel

[Gottfried Haider]

I just filled this as https://bugzilla.gnome.org/show_bug.cgi?id=774370

Tentatively put it into gst-plugins-bad, but it could of course also
be a different component that prevents the playback from working.

Thanks in advance,
Gottfried

On Sat, Nov 12, 2016 at 3:07 PM, Matthew Waters <[hidden email]> wrote:

> On 12/11/16 11:32, Gottfried Haider wrote:
>> Our code [1] is modeled after the testegl example in gst-omx. In 1.8,
>> in handling the "handoff" signal, we would retrieve a (nonzero) tex_id
>> from the GstBuffer we're handed.
>>
>> With 1.10 the code doesn't throw any particular error, the callback is
>> still begin periodically executed - but the tex_id is always zero.
>>
>> Any idea about what could have caused this change in behavior?
>>
>> I am testing on macOS 10.11.6 with the gstreamer-1.0-1.10.0-x86_64
>> binary from the site.
>
> Nothing comes to mind with this little information.
>
> A small reproducible test case would be great to attach to a bug :)
>
> Cheers
> -Matt
>
>> Thanks in advance,
>> Gottfried
>
>
>

_______________________________________________
gstreamer-devel mailing list
[hidden email]
https://lists.freedesktop.org/mailman/listinfo/gstreamer-devel