gstreamer/docs/design/part-TODO.txt
Wim Taymans b491e19e0e docs/design/part-TODO.txt: Add some thoughts and problems with upstream renegotiation.
Original commit message from CVS:
* docs/design/part-TODO.txt:
Add some thoughts and problems with upstream renegotiation.
2008-07-31 13:06:56 +00:00

103 lines
4.8 KiB
Text

API/ABI
-------
- implement return values from events in addition to the gboolean. This should
be done by making the event contain a GstStructure with input/output values,
similar to GstQuery. A typical use case is performing a non-accurate seek to a
keyframe, after the seek you want to get the new stream time that will
actually be used to update the slider bar.
- make gst_pad_push_event() return a GstFlowReturn so that we can resend
NEWSEGMENT and other events.
- GstEvent, GstMessage register like GstFormat or GstQuery.
- query POSITION/DURATION return accuracy. Just a flag or accuracy percentage.
- add some sort of time/frame stepping functionality, either with a flag on
the seek event or some new seek event type. The idea would be to operate on
the current playback position instead of the current configured segment when
doing the seek.
Idea is that frame stepping forwards can be done in the sinks, ie, just
dropping N frames/time, sending more complicated queries upstream which can
ideally handle those cases more efficiently too.
- use | instead of + as divider in serialization of Flags
(gstvalue/gststructure)
- rethink how we handle dynamic replugging wrt segments and other events that
already got pushed and need to be pushed again. Might need GstFlowReturn from
gst_pad_push_event().
- keep track of seeks with a counter so that we can match seek events received
in the demuxer srcpads. This is needed because a normal seek on a pipeline
will send the seek event on all sinks, which results in the demuxer receiving
the seek twice. If there is no way to see that the seek is the same, it will
perform the seek twice.
It would also be nice to have this same sequence number in the segment event
that resulted from the seek so that everything seek related can be tracked
properly.
- Optimize negotiation. We currently do a get_caps() call when we link pads,
which could potentially generate a huge list of caps and all their
combinations, we need to avoid generating these huge lists by generating them
incrementaly when needed. We can do this with a gst_pad_iterate_caps() call.
We also need to incrementally return intersections etc, for this.
- When an element goes to PAUSED there is no way to figure out the running time
when this happened. One could think that we can store this time in the
base_time field of the element but that causes problems when the element is
still using the base_time before really PAUSING. We seem to need a new element
field for this. The running time when an element is paused can be usefull to
clip late buffers instead of prerolling on them.
- Elements in a bin have no clue about the final state of the parent element
since the bin sets the target state on its children in small steps. This
causes problems for elements that like to know the final state (rtspsrc going
to PAUSED or READY is different in that we can avoid sending the useless
PAUSED request).
- Make serialisation of structures more consistent, readable and nicer code-wise.
- When a seekable live source does a flushing seek, it needs a new base_time to
timestamp new data. The pipeline however does not know that there is a live
source in the pipeline and thus does not select a new base_time automatically.
There needs to be a mechanism for a live source to request a new base_time or
pipeline restart.
- pad block has several issues:
* can't block on selected things, like push, pull, pad_alloc, events, ...
* can't check why the block happened. We should also be able to get the item/
reason that blocked the pad.
* it only blocks on datapassing. When EOS, the block never happens but ideally
should because pad block should inform the app when there is no dataflow.
* blocking should only happen from one thread. If one thread does pad_alloc
and another a push, the push might be busy while the block callback is done.
* maybe this name is overloaded. We need to look at some more use cases before
trying to fix this.
- rethink the way we do upstream renegotiation. Currently it's done with
pad_alloc but this has many issues such as only being able to suggest 1 format
and the need to allocate a buffer of this suggested format (some elements such
as capsfilter only know about the format, not the size). We would ideally like
to let upstream renegotiate a new format just like it did when it started.
This could, for example, easily be triggered with a RENEGOTIATE event.
IMPLEMENTATION
--------------
- implement more QOS, see part-qos.txt.
- implement BUFFERSIZE.
- implement pad_block with probes? see above.
DESIGN
------
- unlinking pads in the PAUSED state needs to make sure the stream thread is not
executing code. Can this be done with a flush to unlock all downstream chain
functions? Do we do this automatically or let the app handle this?