commit 144b1c40e9cf08a584c50e1bef7ba3f287e81a4f
Author: Andreas Huber <andih@google.com>
Date: Wed Dec 16 09:28:23 2009 -0800
This H264 file shows a certain problem even better.
commit 3245f1f3b7471975aeeb824a756c987abd610f55
Author: Andreas Huber <andih@google.com>
Date: Wed Dec 16 09:20:08 2009 -0800
Using only the QA testfiles now.
commit 074817eb3816c5dd70858a3594e3b92d799d873b
Author: Andreas Huber <andih@google.com>
Date: Tue Dec 15 16:17:39 2009 -0800
Yay, roles are back again now that the API is in place.
commit 6d847e4932cc38301ae27cb7283b7f1553a95457
Author: Andreas Huber <andih@google.com>
Date: Tue Dec 15 13:01:20 2009 -0800
Added commandline option for specifying the random seed for reproducable tests.
commit 62ab37b26336eaa67e49791c41c996acb6acee3f
Author: Andreas Huber <andih@google.com>
Date: Mon Dec 14 10:53:27 2009 -0800
When issuing a seek it is important that only the first MediaSource::read call has the seek option.
commit e77c46644b2fb6862bafa3569f7d304252074f1e
Author: Andreas Huber <andih@google.com>
Date: Mon Dec 7 16:39:07 2009 -0800
Make sure the tests are actually built, sp<OMXCodec> becomes sp<MediaSource>
commit 6df56915bd55a9445b3c6f953d3cc251d81579b8
Author: Andreas Huber <andih@google.com>
Date: Thu Dec 3 14:25:36 2009 -0800
Temporarily disable support for querying the roles of OMX components.
commit 31bb26930df9e3658dea684cedb4b0f1a06a4a88
Author: Andreas Huber <andih@google.com>
Date: Tue Dec 1 13:36:52 2009 -0800
Disregard EOS events, slightly change the way the EOS flag on output buffers is handled.
commit 4c382fbc9aebee8197d5988d04378062809e7c48
Author: Andreas Huber <andih@google.com>
Date: Tue Dec 1 09:37:24 2009 -0800
New random seek test for the codec tests. Fixed "sticky" end-of-output-buffers flag behaviour in OMXCodec.
commit c762eac3e44309592b61a168d66e091cf609fa03
Author: Andreas Huber <andih@google.com>
Date: Tue Nov 3 14:13:43 2009 -0800
Fix a typo.
commit 50540a59b65c7d476b0193c7494cd75895e6ca6d
Author: Andreas Huber <andih@google.com>
Date: Tue Nov 3 09:48:35 2009 -0800
Some more fine tuning of the unit tests, make MPEG4Extractor less verbose.
commit 1157a7e52a0636706caa235abe16d2ff8a0b8140
Author: Andreas Huber <andih@google.com>
Date: Wed Oct 28 12:01:01 2009 -0700
Changes to the IOMX::listNodes API, this now returns the component's roles as well, unit tests now test all components in all supported roles by default.
commit 30fbf2d8c6cb927689f7ba75eb550a81e9df488a
Author: Andreas Huber <andih@google.com>
Date: Mon Oct 26 09:45:26 2009 -0700
Initial check-in of unit tests for OMX components.
From IrOBEX, 3.3.6.1, we can see below explanation for Flags bit 1:
1 Don't create directory if it does not exist, return an error instead.
The correct understanding should be: if bit 1 is set to 1, we'll not create
directory if it not exist; if bit 1 is set to 0, we'll create it if it not
exists.
This takes care of the conditions where the bluez audio drivers
are not up and we try to make the call to connect the sink.
This would get rid of the hack in Settings app.
Use a NoTitleBar_Fullscreen dialog to implement the
full screen plugin. This runs in the same thread as
WebView (UI in the Browser case). One catch is that
the SurfaceView provided by the plugin needs to be
opaque if it doesn't want to see through the WebView.
The PluginFullScreenHolder translates the events to
the underline WebView. Special treatment in the touch
case as it needs to translate the coordinates.
WebView can't be panned, or double tap to zoom, or
long press to trigger the context menu while having
a full screen plugin.
Inside webkit, we also give the plugin element focus
when it goes to the full screen so that it takes key
events. While handling key events, we don't let it
loose focus or scroll out.
Todo:
When a plugin goes to full screen, we should make
sure the embedded plugin is fully visible. Otherwise
when we translate the touch events back, they will be
outside of the visible rect and will be ignored.
This is part 1 of 2-project check in.
Instead of processing each notification separately event (scheduling an alarm with AlarmManager for each event, posting notification to NotificationManager for each event, etc), only one alarm should be fired for each wake-up time. AlertService will look through the CalendarAlerts table and decides how to update the notication manager. This is more efficient and simplifies the codecode. It is also the long-term fix to elimiate overlapping notification ring tones e.g. DrooDrooDrooooid b/2194249
Also consolidated alarm scheduling code to one place.
Change-Id: If58ce278d55b28abad703f8aeee466d44952fcd0
Pass point to click in three steps.
1) See if point corresponds to cached node.
2) If so,send a message to webkit to see if the node is good.
3) Call webkit with point, and good/bad info.
This is a two-part change with external/webkit.
Fixes http://b/2249425
Fix for http://b/issue?id=2178786
Provide strings for input types file, submit, and reset, and allow
them to be passed down to native.
Requires a change in external/webkit.
Use this in various places where it should serve no purpose to deliver
both broadcasts. This is intended to reduce somewhat the flurry of
broadcasts that we churn through during boot.
* changes:
AwesomePlayer now renders either remote (IOMXRenderer) or locally depending on what kind of decoder was instantiated. Split off color conversion code into its own shared library.
This change fixes the context menu trigger behavior while
the user is selecting via touch. How if a user is selecting
text via dragging their finger, to trigger the context menu
they will have to lift their finger up, then issue a
longpress. This is consistent with the behavior of selecting
via the trackball.
The problem is due to spurious headset connection indications received from event observer when a 3.5mm headset w/o mic is connected.
In this particular case, The HeadsetObserver first received a valid indication of headset with mic connection, followed by a headset with mic disconnection and finally a headset w/o mic connection.
The HeadsetObserver delays the headset disconnection intent to leave time to music app to pause music before the output path is switched.
As the last headset w/o mic connection indication is received from the event observer before the intent corresponding to the spurious headset with mic disconnection is broadcast, the later is discarded. Results a state where the headset with mic is always considered as connected.
The fix consists in not canceling pending intents when a new headset state is received and carrying the HeadsetObserver state with the delayed message triggering the broacast of the disconnection intent.
