| Commit message (Collapse) | Author | Age | Files | Lines |
|\ |
|
| |\
| | |
| | |
| | |
| | |
| | |
| | | |
4.6-integration
* '4.6' of scm.dev.nokia.troll.no:qt/qt-s60-public:
Return correct formats supported lists from Symbian audio backend
|
| | |
| | |
| | |
| | | |
Reviewed-by:Gareth Stockwell
|
| |/
|/|
| |
| |
| |
| |
| | |
Docs stated that the bufferSize was a value in milliseconds.
Corrected this to bytes.
Reviewed-by: trustme
|
|/
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
* QAudioDeviceInfoInternal functions now call CMMFDevSound::InitializeL()
The root cause of QTBUG-10205 was that CMMFDevSound::IntitializeL()
was not called before CMMFDevSound::Capabilities(), resulting in
incorrect values being returned by QAudioDeviceInfo 'supported format'
functions. Resolving this was not straightforward because, while
QAudioDeviceInfo is an entirely synchronous API,
CMMFDevSound::InitializeL() is an asynchronous function. The changes
therefore include a while(QApplication::instance()->processEvents(...))
loop, which is not a particularly elegant solution.
* Encapsulated all interaction with CMMFDevSound API in DevSoundWrapper
class
Because the original bug fix required QAudioDeviceInfoInternal to call
CMMFDevSound::InitializeL(), MDevSoundObserver callback functions had
to be implemented in QAudioDeviceInfoInternal. Rather than pollute
QAudioDeviceInfoInternal, a new class, DevSoundWrapper, was created
which encapsulates all interaction with CMMFDevSound, and therefore
implements MDevSoundObserver. QAudioInputPrivate and
QAudioOutputPrivate now call CMMFDevSound via DevSoundWrapper, with
only the required MDevSoundObserver callbacks forwarded on via signals.
After fixing this bug, running the auto test suites exposed a number
of regressions which had been introduced by the necessary refactoring
described above; this commit therefore also includes fixes for the
following:
* No stateChanged() signal emitted during DevSound initialization
Previously, QAudioInput / QAudioOutput would emit a state change from
StoppedState to IdleState or ActiveState, as soon as start() was called.
In the case where the requested format is subsequently found to be
unsupported, in addition to the error() value being set to OpenError,
a second state change would be emitted, back to StoppedState. This is
not the behaviour exhibited by the other platforms' backends,
and therefore caused an auto test failure. Now, the backend only
transitions to IdleState / ActiveState on successful initialization.
* Stop emitting notify() signal when notifyInterval is set to zero
* Call CMMFDevSound::RecordInitL() from QAudioInput::resume() if no
buffer is currently held
This is necessary in order to restart the data flow.
* Call CMMFDevSound::BufferFilled() from QAudioOutput::resume() in push
mode
The auto test does not resume pushing data to QAudioOutput until
'bytesFree() >= periodSize()' becomes true. Because, for the Symbian
backend, periodSize() == bufferSize(), this condition is only met when a
new buffer is provided by DevSound via BufferToBeFilled().
Task-number: QTBUG-10205
|
|
|
|
|
|
|
|
| |
- fix deadlock with QAudioInput (win32) backend
- changed win32 backend to use QMutex lock
- setNotifyInterval(0) to disable notify() signal
Reviewed-by:Derick Hawcroft
|
|
|
|
| |
Reviewed-by:Derick Hawcroft
|
|
|
|
|
| |
Task-number:QTBUG-9766
Reviewed-by:Derick Hawcroft
|
|\
| |
| |
| |
| |
| |
| | |
4.6-integration
* '4.6' of scm.dev.nokia.troll.no:qt/qt-multimedia-staging:
Bug in QAudioOutput: Invalid use of "waveOutUnprepareHeader"
|
| |
| |
| |
| |
| |
| |
| | |
Fixed freeBlocks func in both input and output
Task-number:QTBUG-9875
Reviewed-by:Andrew den Exter
|
| |
| |
| |
| | |
Reviewed-by: trustme
|
|/
|
|
|
|
|
|
|
| |
Work done jointly by Gareth and me. Yields no qdoc errors.
Task-number: QTBUG-9342
Task-number: QTBUG-9120
Reviewed-by: Gareth Stockwell
Reviewed-by: David Boddie
|
|
|
|
|
|
|
|
| |
Increased default buffer size and period size for windows.
11025monoS16LE was choppy, needed bigger periods.
Task-number:QTBUG-9101
Reviewed-by:Derick Hawcroft
|
|
|
|
|
|
|
| |
record from different soundcards
Task-number:QTBUG-9221
Reviewed-by:Andrew den Exter
|
|
|
|
|
|
|
|
|
| |
QtMultimedia backend for Windows
-Added 8000 as long as another frequency is available.
-options available changed from hard-coded to detected options.
Task-number:QTBUG-9100
Reviewed-by:Derick Hawcroft
|
|
|
|
|
|
|
|
| |
Removed state change from close() function, should be set by caller
depending on context.
Task-number:QTBUG-9357
Reviewed-by:Dmytro Poplavskiy
|
|\
| |
| |
| |
| |
| |
| | |
4.6-integration
* '4.6' of scm.dev.nokia.troll.no:qt/qt-multimedia-staging:
Calculate period size correctly.
|
| |
| |
| |
| | |
Reviewed-by:Dmytro Poplavskiy
|
|/
|
|
|
|
|
|
|
|
|
|
|
| |
Because of the logic in the Symbian implementations of suspend()
and processedUSecs(), the behaviour prior to this patch was as follows:
- after calling suspend(), processedUSecs() reported zero
- when resume() is called, the amount of silent padding data inserted into
the output stream is equal to the amount of data played prior to suspend()
The patch corrects the above defect, but introduces a new one, which is
described in QTBUG-9322
Reviewed-by: trustme
|
|\
| |
| |
| |
| |
| |
| |
| |
| | |
4.6-integration
* '4.6' of scm.dev.nokia.troll.no:qt/qt-multimedia-staging:
Use AudioConverter when not preferred format.
Handle values passed to setNotifyInterval more robustly.
Make sure to check for a valid audioformat before open.
|
| |
| |
| |
| | |
Reviewed-by:Dmytro Poplavskiy
|
| |
| |
| |
| | |
Reviewed-by:Dmytro Poplavskiy
|
| |
| |
| |
| | |
Reviewed-by:Dmytro Poplavskiy
|
|/
|
|
|
|
|
|
|
| |
Changed backend from being a plugin, to being statically compiled into
QtMultimedia.dll. This is to achieve consistency with the backends for
other plugins, and to allow the plugin API to be modified.
Task-number: QTBUG-8994
Reviewed-by: trustme
|
|\ |
|
| |\
| | |
| | |
| | | |
into 4.6
|
| | |
| | |
| | |
| | | |
Reviewed-by:Dmytro Poplavskiy
|
| | |
| | |
| | |
| | | |
Reviewed-by:Dmytro Poplavskiy
|
| |/
|/|
| |
| |
| |
| |
| | |
Handle error recovery better.
Task-number:QTBUG-8893
Reviewed-by:Derick Hawcroft
|
|/
|
|
|
|
|
|
| |
Bug introduced by change 04532ba052559b265b1bc85dc143d8aeeb02149f
Was calculating time assuming totalTimeValue was in frames but it
is stored as bytes in alsa input case.
Reviewed-by:Dmytro Poplavskiy
|
|
|
|
|
| |
Task-number:QTBUG-8755
Reviewed-by:Derick Hawcroft
|
|
|
|
|
|
|
| |
83e73a69d5a1c6b94c5c83f91141f35d65d27f36.
alsa callback was deadlocking with QEventDispatcherGlib.
Reviewed-by:Dmytro Poplavskiy
|
|
|
|
| |
Reviewed-by:Dmytro Poplavskiy
|
|
|
|
|
|
|
|
| |
Adjust buffer and period sizes if system cannot work with defaults
(alsa).
Task-number:QTBUG-8339
Reviewed-by:Derick Hawcroft
|
|
|
|
| |
Reviewed-by: Kurt Korbatits
|
|
|
|
|
| |
Task-number: QTBUG-5741
Reviewed-by: Kurt Korbatits
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
New unit tests have identified issues with awin32 backend. These issues
need to be resolved before new unit tests are added.
* stateChanged() signal should only be emitted on change of state
currently signals can be emitted multiple times.
* for output data is being lost when input can be read from
QIODevice but cannot be written to audio sub system.
Task-number:QTBUG-8441
Reviewed-by:Dmytro Poplavskiy
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
New unit tests have identified issues with alsa backend. These issues
need to be resolved before new unit tests are added.
* stateChanged() signal should only be emitted on change of state
currently signals can be emitted multiple times.
* elapsedUSecs() currently uses alsa to provide this value but
this is not reliable enough and this time is reset when
suspend/resuming. This is not correct operation.
* for output data is being lost when input cant be read
from QIODevice but cannot be written to audio sub system.
Task-number:QTBUG-8440
Reviewed-by:Dmytro Poplavskiy
|
|
|
|
|
|
|
| |
Detail state changes and error states in QAudioInput and QAudioOutput
documentation.
Reviewed-by: Derick Hawcroft
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
This reverts commit 80d4a4945d3273a4b2ce91e34597533f661af320.
Conflicts:
examples/multimedia/audioinput/audioinput.cpp
examples/multimedia/audiooutput/audiooutput.cpp
src/multimedia/audio/qaudio_mac.cpp
src/multimedia/audio/qaudiodeviceinfo_alsa_p.cpp
src/multimedia/audio/qaudiodeviceinfo_mac_p.cpp
src/multimedia/audio/qaudiodeviceinfo_win32_p.cpp
src/multimedia/audio/qaudioformat.h
src/multimedia/audio/qaudioinput.cpp
src/multimedia/audio/qaudiooutput.cpp
tests/auto/qaudiodeviceinfo/tst_qaudiodeviceinfo.cpp
tests/auto/qaudioformat/tst_qaudioformat.cpp
tests/auto/qaudioinput/tst_qaudioinput.cpp
tests/auto/qaudiooutput/tst_qaudiooutput.cpp
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
This reverts commit f124538ef4840c3d24b4c7e9e7221adb52bdee2c.
Conflicts:
examples/multimedia/audioinput/audioinput.cpp
examples/multimedia/audiooutput/audiooutput.cpp
src/multimedia/audio/qaudio_mac.cpp
src/multimedia/audio/qaudiodeviceinfo_alsa_p.cpp
src/multimedia/audio/qaudiodeviceinfo_mac_p.cpp
src/multimedia/audio/qaudiodeviceinfo_win32_p.cpp
src/multimedia/audio/qaudioformat.h
src/multimedia/audio/qaudioinput.cpp
src/multimedia/audio/qaudiooutput.cpp
tests/auto/qaudiodeviceinfo/tst_qaudiodeviceinfo.cpp
tests/auto/qaudioformat/tst_qaudioformat.cpp
tests/auto/qaudioinput/tst_qaudioinput.cpp
tests/auto/qaudiooutput/tst_qaudiooutput.cpp
|
|\
| |
| |
| |
| |
| |
| | |
4.6-integration
* '4.6' of scm.dev.nokia.troll.no:qt/qt-multimedia-team:
Fix example in QAudioOutput/QAudioInput docs
|
| |
| |
| |
| |
| |
| | |
Fixed error in doc comments.
Reviewed-by:Justin McPherson
|
| |
| |
| |
| |
| |
| |
| |
| |
| | |
The example given for QAudioOutput doesn't actually compile:
QAudioOutput::start takes a pointer to a QIODevice, but it's passed
a QFile (not the pointer).
Merge-request: 418
Reviewed-by: Thiago Macieira <thiago.macieira@nokia.com>
|
|\ \
| |/
| |
| |
| | |
Conflicts:
src/multimedia/audio/qaudioformat.cpp
|
| |
| |
| |
| |
| |
| |
| |
| | |
The alsa backend was calculating sample timer on resume using buffer_time
instead of period_time.
Task-number:QTBUG-7044
Reviewed-by:Justin McPherson
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
-Added channelCount(), changed everything to use this instead
of channels() in QAudioFormat class.
-Added setSampleRate() and sampleRate(), changed everthing to use these
instead of setFrequency() and frequency() in QAudioFormat class.
-Added supportedSampleRates() and supportedChannelCounts(), changed
everything to use these instead of supportedFrequencies() and
supportedChannels() in QAudioDeviceInfo class.
Reviewed-by:Justin McPherson
|
| |
| |
| |
| |
| |
| |
| | |
Added setChannelCount() and updated docs/examples/tests to use it
instead of setChannels().
Reviewed-by:Justin McPherson
|
| |
| |
| |
| |
| |
| | |
Also added it to the multimedia group.
Reviewed-by: Trust Me
|