* This is the only solution that allowed to use the best
of both ways to do this calculus. I've also tested it
with a modified sound player that snoozed every time
the buffer should be handled, and found that neither
of the lateness calculus I tested (including enqueue_time)
really solve all problems. That's why I've tried to find
an average solution. There's still room for improvements
eventually.
* Move !missing_symbols.cpp in the legacy directory.
* Move useless methods in a Deprecated.cpp file.
* SetRealtimeFlags / GetRealtimeFlags are useless, they
don't make sense nowadays.
* The symbols are anyway preserved.
* People interested, please review!
* This is based on the patch from Fredrik Moden which was based on
the Oleg Krysenkov one.
* The original patch has been reworked by myself.
* Adapted the code to work with the new PluginManager API which
differently than before doesn't need to contact the media_server.
* The first problem was the O(n^2) complexity of the algorithm, it's
now linear and try to act in a circular way by dispatching
events and reading the port in a balanced way. This exclude
a certain degree of possible deadlocks.
* Add detection and escape when the system try to kill the
thread. This solve some blocking issues on exit et similia
that i had with libjackcompat.
* The algorithm choose soon which event to focus on.
* Lateness is calculated just before the event is dispatched
as it is the more appropriate place, otherwise we would be
calculating something imprecise/guessed.
* Remove timed_event_queue::queued_time. It's more precise to
just use the RealTime() before to Dispatch the event.
* It should solve the BSoundPlayer lateness problems.
* With those improvements the media_kit is not going to lock
completely under stress conditions, instead it try to work
in a best effort shape.
* There's still room for improvements, for example i'm considering some
strategies in lateness situations such as update scheduling latency,
try to decrease waiting time and detect when we are too early on
the other hand to recover when the load go down.
* Thanks to Julian Harnath for sharing his WIP patch which helped
with some controls such as avoiding negative lateness.
* Comments are welcome!
* Due to the needs to provide a peaceful UX, i moved out the
notifications mechanism from the Media preflet and i have
integrated it with the launch and shutdown functions of MediaDefs.h.
* This implied to introduce a new launch_media_server function
similar to the shutdown_media_server allowing to specify a custom
notification function too.
* Both functions then are reworked to send by default notifications
to the Deskbar, this was needed because in a lot of situations
the mechanism failed without correctly noticing the user.
* The one argument launch_media_server is considered to be deprecated,
the default argument is removed to mantain binary compatibility but
make new apps to automatically use the new one with just the default
arguments. This is needed due to conflicts in overloading.
* Improve notifications by indirectly extending localization
to old BeOS apps.
* This is done by watching to registrar notifications
and providing a minimal service to contact the
media roster in private API. The roster use this
service to automatically reconnect to the media_server.
- This allows a BMediaDecoder (e.g. FFMPEG Plugin) to communicate back format
changes to its clients.
For a more thorough explanation and discussion see haiku-development mailing
list: http://www.freelists.org/post/haiku-development/Request-for-protest-Media-Kit-Extend-media-header-struct
- Backwards compatibility is taken into account and preserved by reducing the
relevant _reserved[] fields.
- Code changes that will actually make use of these extensions are due for the
following commits. As these structure extensions affect several Haiku
components (e.g. media_server, MediaPlayer, etc.) and third party apps (e.g.
StampTV, etc.) I refrain from committing them in one batch with this commit.
This should make it easier to track down bugs originating in this code
change.
(cherry picked from commit 806b6888d2dcf84b4934f8f137a48d3381864d1c)
that they can modify the media_format passed in. For example they
can store information in the user_data section. I don't actually
use this anymore, but it may come in handy again.
AVFormatWriter:
* Adjust the AVCodecContext flags not only for video, but also
for audio streams (as the API example does). This mechanism
may not yet work, since the AVCodecEncoder actually uses a
different AVCodecContext instance.
* Use the encodeInfo->flags and specify the key frame flag
for the AVPacket. This finally makes videos encoded on Haiku
seekable.
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@39035 a95241bf-73f2-0310-859d-f6bbb57e9c96
BMediaFiles and about BMediaTracks in BMessages. As an
example, one can get chapter meta-data or the language
name of an audio-track.
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@38685 a95241bf-73f2-0310-859d-f6bbb57e9c96
BBufferConsumer::BufferReceived() cannot know whom to send the "buffer is
late" notification (unless we only have a single input). To solve this, the
media_header now contains extra fields that can be used to create a
media_source object.
* Unfortunately, BBufferProducer::SendBuffer() cannot know the output either in
case there is more than one. Hence, I deprecated the existing SendBuffer()
call and moved it into "private" - IOW old sources using it won't compile
anymore under Haiku.
* I introduced a new SendBuffer() variant that also gets the media_source as
argument.
* Updated all sources (that are part of the image) to use the new variant.
* Removed some purposely commented out code in the audio mixer.
* Implemented late buffer notification, as well as late buffer handling in the
audio mixer; this is a bit of work in progress, so the debug output is left
in there.
* Some cleanup.
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@36184 a95241bf-73f2-0310-859d-f6bbb57e9c96
* Small coding style cleanups in the .cpp
* When reading raw chunks, keep fCurrentFrame updated anyway
* When decoding frames and chunks, don't change the meaning of fCurrentTime,
it's supposed to be the start time of the next chunk/frame, same as
CurrentFrame(), not the time of the frame/chunk we just decoded/read.
BMediaTrack::CurrentFrame() is actually documented like this in the BeBook,
but CurrentTime() is not. However, when seeking to a specific time, it is
understood that this is the time of the next frame/chunk. If we decode/read
it and set fCurrentTime to the start_time as specified in the media_header,
it would actually not change for the first frame/chunk after seeking.
To be able to know the duration of chunks, fWriterFormat had to be changed
to just fFormat and is now used for decoding as well.
* Simplified handling of media_header in some methods, saves one assignment.
These changes are not so well tested, yet.
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@34923 a95241bf-73f2-0310-859d-f6bbb57e9c96
look into this now.
* Renamed the MMediaFilesManager to MediaFilesManager.
* Got rid of its Map usage, we're using std::map now instead.
* The media_server is using an unsafe method to transfer areas from the server
to the client. I've changed this for the BMediaFiles API to use Haiku's new
transfer_area() function. However, this seems to cause the above mentioned
problems.
* Improved naming and data exchange structures used for the BMediaFiles API.
* Fixed BMediaFiles leaking its lists.
* Cleanup.
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@34554 a95241bf-73f2-0310-859d-f6bbb57e9c96
into smaller parts:
media_addon_server:
* Removed (broken) use of (broken and inefficient) home-brewn Map, and List
classes. This also fixes a crash on shutdown when used with the malloc_debug
implementation. It's using stl::vector, and stl::map now instead.
_shared_buffer_list:
* Renamed _shared_buffer_list to SharedBufferList, and put it into the BPrivate
namespace. Also, made a class out of it.
* Separated shared buffer list creation from cloning.
* Enlarged maximum number of buffers to something that is not that evil, but
actually uses the space it has (ie. is a useful multiple of
shared_buffer_info that fills a multiple of B_PAGE_SIZE as much as possible).
* No longer drops into the debugger if the
* The list that is currently used is very inefficient for the features it
provides though (no change there).
_buffer_id_cache:
* Renamed to BufferCache, and put it into the private namespace
* It now deletes its buffers on deletion; since the BBufferConsumer will be
gone, too, at this point, there is little chance that there are still buffers
in use.
* Also, it's now using std::map instead of the (see above) Map class.
BBuffer:
* Got rid of the fBufferID member.
Misc.:
* Got rid of the global "team" variable; the media kit is now using the
private app kit's current_team() now.
* Added a lot of missing error checks (mostly memory allocations).
* Renamed fields like "flavorid" to flavor_id, renamed "dfi_*" fields to
something more detailed.
* Moved ServerInterface.h from src/servers/media/ to headers/private/media.
* Notifications.h was not self contained.
* Added missing licenses.
* Lots of cleanups, and coding style fixes.
What this doesn't fix:
* Bug #4954 which started all this (this comes next, though)
* Deinitialization is broken, as the PortPool is uninitialized too early, and
still used afterwards.
* The strange add-on monitoring code in the media_addon_server
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@34500 a95241bf-73f2-0310-859d-f6bbb57e9c96
loader's heap. Unlike the BeBook documents, we always lock their memory,
though.
* Added Haiku extension rtm_available() that returns how much space is left in
a pool.
* I've disabled the undocumented functions for now - please open a bug report
if you encounter them used in applications.
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@34419 a95241bf-73f2-0310-859d-f6bbb57e9c96
* Removed private types from MediaEncoder.h (These have Haiku counterparts,
I expect MediaEncoder is not yet implemented...)
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@32787 a95241bf-73f2-0310-859d-f6bbb57e9c96
* Cleaned up the source file, no functional change (intended).
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@32139 a95241bf-73f2-0310-859d-f6bbb57e9c96
* Improved Encoder API towards what we need for the get_next_encoder() variants
and the BMediaTrack API.
* Implemented the rest of MediaWriter. Still undecided what to make of
AddTrackInfo(). BMediaEncoder has that as well, which hints that this is
something the Encoder needs to support. But it could also be that this is
only possible to support in Writer.
* Wired a lot of previously unimplemented methods in BMediaFile and BMediaTrack
needed for write support. If I have not overlooked anything, only the
parameter stuff is still unimplemented now.
This is all untested, since the FFMpeg Encoder and Writer are still only stubs.
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@32013 a95241bf-73f2-0310-859d-f6bbb57e9c96
* Removed the BParameterWeb::{Start|Stop}Watching() methods again, as the same
thing can be achieved using the BMediaRoster already (even though it doesn't
seem to work yet).
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@30058 a95241bf-73f2-0310-859d-f6bbb57e9c96
stubs.
* However, without a working Dano at hand, I can't really implement something
compatible, so I'll postpone this to some later point.
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@30030 a95241bf-73f2-0310-859d-f6bbb57e9c96
* Adapted and cleaned up ParameterWeb.cpp.
* Fixed some copy&paste bugs in the BTextParameter.
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@30029 a95241bf-73f2-0310-859d-f6bbb57e9c96
