+ Make GESAssetFileSource a subclass of it
+ Remove ges_asset_filesource_get_supported_type as it is now in GESAssetTimelineObject
+ Remove the GES_META_TIMELINE_OBJECT_SUPPORTED_FORMATS as it is useless now
Modifies some API:
ges_timeline_object_create_track_objects now take a GESTrackType instead of a
GESTrack as second argument, and return a GList instead of a boolean
ges_timeline_object_create_track_object now take a GESTrackType instead of a
GESTrack as second argument
+ Avoid to assume function arguments are correct before actually testing
them in ges_timeline_object_add_track_object
API: ges_timeline_object_add_asset
We used to do it in TimelineFileSource which does not make sense.
At the same time we set AUDIO | VIDEO as default supported types as it is more
likely to be what subclasses support. If it is not the case, they need to
specify it as shown in ges-timeline-custom-timeline-source.c
+ Fix the tests accordingly
+ timeline: Add a snapping-distance property
+ Bump the GLib dependency to 2.28 in the mean time as we need some functions from GSequence that only landed
+ Update the testsuite accordingly
API: GESTimeline:snapping-distance property
API: ges_timeline_object_edit
API: ges_timeline_object_ripple
API: ges_timeline_object_ripple_end
API: ges_timeline_object_roll_start
API: ges_timeline_object_roll_end
API: ges_timeline_object_trim_start
API: ges_track_object_edit
API: GESEdge enum
API: GESEditMode enum
This makes more sense to be a guint64 as it actually is a GstClockTime,
and this way we keep the API concistent.
(This code has not been release so we can still change the API.)
This property was firstly added to TrackFileSource, but in the end, it makes
more sense for it to be directly in TrackOject as it can be usefull in other cases.
ges_timeline_object_add_track_object actually calls
ges_track_object_set_timeline_object so do not do it once more ourself.
Especially since it results in having a TrackObject.timeline_object refering
to a TimelineObject it is not actually in yet.
We were emitting it before it gets added to the track_object list,
so the list we were getting with get_top_effects was containing
a TrackObject that wasen't a TrackEffect
+ A bit of refactoring
Record the TrackObject that initiated a TimelineObject movement so we don't
get inifite loops.
Also fix the new TrackObject calculation:
child.start = time - offset (not time + offset)
