GitLab

When pushing the remaining data on EOS, don't just push whatever
data is left in the adapter, but only push data that's at least
of min-size.

Don't try to push remaining data in the adapter on receiving a FLUSH event,
just flush the adapter. Do this on FLUSH_STOP, however, which is serialized,
unlike FLUSH_START, so we don't mess with the adapter at the same time as
the streaming thread.

https://bugzilla.gnome.org/show_bug.cgi?id=649005

https://bugzilla.gnome.org/show_bug.cgi?id=648929

Avoids a massive leak :)

https://bugzilla.gnome.org/show_bug.cgi?id=648929

https://bugzilla.gnome.org/show_bug.cgi?id=648929

This can happen with AFC-only packets. Avoids leaking buffers.

https://bugzilla.gnome.org/show_bug.cgi?id=648929

It doesn't seem to work in an uninstalled setup, and
breaks make distcheck for me.

Rename the list and dist them. We need them for the actual templates.

The template contains things we already define by default.

Use the object class and not the object in the init function. Set the vmethods.
Add default returns.

Move not working templates to a separate variable to highlight the fact that
they need more work. These need at least the class and type fields filled.

Add an easy way to check the element-maker templates.

Get the dirname for the script and use that to reference the templates. Use the
templatedir variable to check for templates.

This allows using capitalized acronyms in class names, so using
"AVC_src" on the command line will create filename gstavcsrc.c,
class name GstAVCSrc, and symbol names gst_avc_src_*.

Fixes #606662.

Fixes #647852.

Both basevideoencoder and basevideocodec were setting
system_frame_number, leading to confusion.  Fixes #647853.

Fixes #647854.

https://bugzilla.gnome.org/show_bug.cgi?id=647364

From c3cafe1 to 46dfcea

https://bugzilla.gnome.org/show_bug.cgi?id=641796

And remove empty and unused variables.