gst/videoscale/gstvideoscale.c: Events have a perfectly good copy_on_write implementation, use it so we preserve GstD...

Original commit message from CVS:
* gst/videoscale/gstvideoscale.c:
(gst_videoscale_handle_src_event):
Events have a perfectly good copy_on_write implementation,
use it so we preserve GstData flags too.

* sys/ximage/ximagesink.c: (gst_ximagesink_ximage_put),
(gst_ximagesink_handle_event), (gst_ximagesink_chain),
(gst_ximagesink_send_pending_navigation), (gst_ximagesink_expose):
* sys/ximage/ximagesink.h:
* sys/xvimage/xvimagesink.c: (gst_xvimagesink_xvimage_put),
(gst_xvimagesink_handle_event), (gst_xvimagesink_chain),
(gst_xvimagesink_buffer_free), (gst_xvimagesink_buffer_alloc),
(gst_xvimagesink_interface_supported),
(gst_xvimagesink_send_pending_navigation),
(gst_xvimagesink_navigation_send_event),
(gst_xvimagesink_set_xwindow_id), (gst_xvimagesink_expose),
(gst_xvimagesink_colorbalance_list_channels),
(gst_xvimagesink_colorbalance_set_value),
(gst_xvimagesink_colorbalance_get_value),
(gst_xvimagesink_set_property), (gst_xvimagesink_get_property),
(gst_xvimagesink_finalize), (gst_xvimagesink_init),
(gst_xvimagesink_base_init), (gst_xvimagesink_class_init),
(gst_xvimagesink_get_type), (plugin_init):
* sys/xvimage/xvimagesink.h:
Move send_pending_navigation functions outside the stream_lock and let
them lock/unlock when they need so we don't hold the stream lock
while sending the event upstream.
Add need_expose flag to avoid an unnecessary screen update if an
incoming buffer beat us to it.
Unset the GST_EVENT_COMMON_FLAG_NEED_RESPONSE flag to avoid blocking on
queues while sending navigation events, since we don't need a reply.
Handle filler events for still frames.

0 files changed, 0 insertions, 0 deletions