Bug 1307820 - 1a. Move GeckoApp EventDispatcher to GeckoView; r=snorp
Make it a GeckoView-specific EventDispatcher instead of
GeckoApp-specific, so that GeckoView consumers can benefit from a
per-view EventDispatcher. In addition, a few events like Gecko:Ready are
moved back to the global EventDispatcher because that makes more sense.
Bug 1307820 - 1b. Don't use GeckoApp EventDispatcher during inflation; r=snorp
During layout inflation, we don't yet have GeckoView and therefore the
GeckoView EventDispatcher, so we should not register events until later,
typically during onAttachedToWindow.
Bug 1307820 - 2. Introduce GeckoBundle; r=snorp
The Android Bundle class has several disadvantages when used for holding
structured data from JS.
The most obvious one is the differentiation between int and double,
which doesn't exist in JS. So when a JS number is converted to either a
Bundle int or double, we run the risk of making a wrong conversion,
resulting in a type mismatch exception when Java uses the Bundle. This
extends to number arrays from JS.
There is one more gotcha when using arrays. When we receive an empty
array from JS, there is no way for us to determine the type of the
array, because even empty arrays in Java have types. We are forced to
pick an arbitrary type like boolean[], which can easily result in a type
mismatch exception when using the array on the Java side.
In addition, Bundle is fairly cumbersome, and we cannot access the inner
structures of Bundle from Java or JNI, making it harder to use.
With these factors in mind, this patch introduces GeckoBundle as a
better choice for Gecko/Java communication. It is almost fully
API-compatible with the Android Bundle; only the Bundle array methods
are different. It resolves the numbers problem by performing conversions
if necessary, and it is a lot more lightweight than Bundle.
Bug 1307820 - 3. Convert BundleEventListener to use GeckoBundle; r=snorp
Convert BundleEventListener from using Bundle to using GeckoBundle.
Because NativeJSContainer still only supports Bundle, we do an extra
conversion when sending Bundle messages, but eventually, as we eliminate
the use of NativeJSContainer, that will go away as well.
Bug 1307820 - 4. Introduce EventDispatcher interfaces; r=snorp
Introduce several new XPCOM interfaces for the new EventDispatcher API,
these interfaces are mostly mirrored after their Java counterparts.
* nsIAndroidEventDispatcher is the main interface for
registering/unregistering listeners and for dispatching events from
JS/C++.
* nsIAndroidEventListener is the interface that JS/C++ clients implement
to receive events.
* nsIAndroidEventCallback is the interface that JS/C++ clients implement
to receive responses from dispatched events.
* nsIAndroidView is the new interface that every window receives
that is specific to the window/GeckoView pair. It is passed to chrome
scripts through window arguments.
Bug 1307820 - 5. Remove EventDispatcher references from gfx code; r=snorp
EventDispatcher was used for JPZC, but NPZC doesn't use it anymore.
Bug 1307820 - 6. General JNI template improvements; r=snorp
This patch includes several improvements to the JNI templates.
* Context::RawClassRef is removed to avoid misuse, as Context::ClassRef
should be used instead.
* Fix a compile error, in certain usages, in the DisposeNative overload
in NativeStub.
* Add Ref::IsInstanceOf and Context::IsInstanceOf to mirror the
JNIEnv::IsInstanceOf call.
* Add Ref::operator* and Context::operator* to provide an easy way to
get a Context object.
* Add built-in declarations for boxed Java objects (e.g. Boolean,
Integer, etc).
* Add ObjectArray::New for creating new object arrays of specific types.
* Add lvalue qualifiers to LocalRef::operator= and GlobalRef::operator=,
to prevent accidentally assigning to rvalues. (e.g.
`objectArray->GetElement(0) = newObject;`, which won't work as intended.)
Bug 1307820 - 7. Support ownership through RefPtr for native JNI objects; r=snorp
In addition to direct ownership and weak pointer ownership, add a third
ownership model where a native JNI object owns a RefPtr that holds a
strong reference to the actual C++ object. This ownership model works
well with ref-counted objects such as XPCOM objects, and is activated
through the presence of public members AddRef() and Release() in the C++
object.
Bug 1307820 - 8. Implement Gecko-side EventDispatcher; r=snorp
Add a skeletal implementation of EventDispatcher on the Gecko side.
Each widget::EventDispatcher will be associated with a Java
EventDispatcher, so events can be dispatched from Gecko to Java and vice
versa. AndroidBridge and nsWindow will implement
nsIAndroidEventDispatcher through widget::EventDispatcher.
Other patches will add more complete functionality such as
GeckoBundle/JSObject translation and support for callbacks.
Bug 1307820 - 9. Implement dispatching between Gecko/Java; r=snorp
Implement translation between JSObject and GeckoBundle, and use that for
dispatching events from Gecko to Java and vice versa.
Bug 1307820 - 10. Implement callback support; r=snorp
Implement callback support for both Gecko-to-Java events and
Java-to-Gecko events.
For Gecko-to-Java, we translate nsIAndroidEventCallback to a Java
EventCallback through NativeCallbackDelegate and pass it to the Java
listener.
For Java-to-Gecko, we translate EventCallback to a
nsIAndroidEventCallback through JavaCallbackDelegate and pass it to the
Gecko listener. There is another JavaCallbackDelegate on the Java side
that redirects the callback to a particular thread. For example, if the
event was dispatched from the UI thread, we make sure the callback
happens on the UI thread as well.
Bug 1307820 - 11. Add BundleEventListener support for Gecko thread; r=snorp
Add support for BundleEventListener on the Gecko thread, so that we can
use it to replace any existing GeckoEventListener or NativeEventListener
implementations that require the listener be run synchronously on the
Gecko thread.
Bug 1307820 - 12. Add global EventDispatcher in AndroidBridge; r=snorp
Add an instance of EventDispatcher to AndroidBridge to act as a global
event dispatcher.
Bug 1307820 - 13. Add per-nsWindow EventDispatcher; r=snorp
Add an instance of EventDispatcher to each nsWindow through an
AndroidView object, which implements nsIAndroidView. The nsIAndroidView
is passed to the chrome script through the window argument when opening
the window.
Bug 1307820 - 14. Update auto-generated bindings; r=me
Bug 1307820 - 15. Update testEventDispatcher; r=snorp
Update testEventDispatcher to include new functionalities in
EventDisptcher.
* Add tests for dispatching events to UI/background thread through
nsIAndroidEventDispatcher::dispatch.
* Add tests for dispatching events to UI/background thread through
EventDispatcher.dispatch.
* Add tests for dispatching events to Gecko thread through
EventDispatcher.dispatch.
Each kind of test exercises both the global EventDispatcher through
EventDispatcher.getInstance() and the per-GeckoView EventDispatcher
through GeckoApp.getEventDispatcher().
Add jni::IsFennec() that returns whether we're in a Fennec environment
(defined as the presence of the GeckoApp class). Then, add
jni::IsFennec() checks to places where we use JNI for Fennec-only classes.
Right now we report Java exception stacks through two code paths:
- GeckoAppShell.reportJavaCrash for exceptions caught by Java code
- jni::HandleUncaughtException for exceptions caught by native code
This patch combines the code for reporting the Java stack trace into
ReportException, to make it easier to add new features such as reporting
OOM exceptions.
We used to only handle certain exceptions in native code, and otherwise
forward the exception to Java handling code. Handling exceptions in
native code gives us better native stack traces in crash reports, so
this patch makes us always handle uncaught Java exceptions in native
code if possible, and only forward to Java as a fallback.
Implement checking the calling thread of a JNI call based on the
calledFrom attribute set in WrapForJNI. Also implement automatic call
dispatching based on the dispatchTo attribute set in WrapForJNI. This
eliminates the use of UsesNativeCallProxy and UsesGeckoThreadProxy.
Improve the performance of JNI calls by making JNI calls require a
Context object. LocalRef inherits from Context and can make calls
directly. Non-local Ref classes will generate a Context object when
making a call. The patch also makes the template design cleaner in
several cases.
This patch adds specializations of jni::Ref for primitive arrays like
IntArray and FloatArray, so that the arrays can be accessed through the
GetElement, GetElements, and Length members.
When we have a Java exception in native code, the Java stack in the
exception will not be very useful because the top frame is the native
entry point. In this case, the native stack is more useful. However,
currently we don't get a good native stack in this situation because we
go through Java when handling the exception, and the native stack we get
will have a lot of unknown frames inside libdvm or libart. This patch
makes us stay in native code when handling an uncaught exception from
native code, so that we get a good native stack.
Add a direct ownership model where the Java object owns the
corresponding C++ object directly, in addition to the WeakPtr model
where the Java object owns a WeakPtr to the C++ object. The WeakPtr
model is chosen when the implementing C++ class inherits from
SupportsWeakPtr. Otherwise, the direct ownership model is chosen. Under
the direct ownership model, a UniquePtr object must be used to attach
the containing C++ object to a Java object, to ensure ownership is
passed on to the Java object.
This patch adds:
* Conversion operator from String::Param to String::LocalRef.
* More overloads of the jni::ThrowException function.
* name members to built-in types like jni::Object, jni::String, etc.
This allows using jni::Accessors::EnsureClassRef on built-in types
to get built-in class refs (e.g. jclass for java/lang/String).
* Ability to implicitly convert LocalRef<Cls> to LocalRef<Object>
* Fixes for bugs in LocalRef/GlobalRef where new refs are not created.
* Fixes for inaccurate uses of mozilla::Forward in favor of mozilla::Move
We use Ref::From() inside TypeAdapter<Ref>::ToNative to convert a raw JNI
ref argument to a Ref argument for the C++ function. However, that
generates a compile error, unless we make TypeAdapter<Ref> a friend of
Ref, because we intentionally made Ref's copy constructor private and
returning from TypeAdapter<Ref>::ToNative requires the copy constructor.
Introduce a JNIObject class that serves as a base class for classes
that wish to use per-instance native methods. JNIObject includes a long
native pointer field that the C++ code accesses to associate the Java
object instance with a C++ object instance.
Jim Chen
Carsten "Tomcat" Book
Nathan Froyd