On Mon, 18 Jan 2021 16:08:37 GMT, Phil Race <p...@openjdk.org> wrote:
> This removes desktop module usage of the JNF JNI reference convenience APIs
> These are simply a direct conversion
> JNFNewGlobalRef
> JNFDeleteGlobalRef
> JNFNewWeakGlobalRef
> JNFDeleteWeakGlobalRef
>
> These two
> JNFJObjectWrapper
> JNFWeakJObjectWrapper
> exist to allow clean up of the refs when a Cocoa wrapper object is released.
> However in all cases there are more direct ways to clean it up and in at
> least one usage
> the existing code directly releases it with the comment that this is more
> efficient.
>
> All our automated regression and JCK tests pass with this change.
src/java.desktop/macosx/native/libawt_lwawt/awt/AWTWindow.h line 55:
> 53: @property (nonatomic, retain) NSWindow *nsWindow;
> 54:
> 55: @property (nonatomic) jobject javaPlatformWindow;
I think it will be useful to have a comment here that this value is a weak
reference and should be copied to the local ref before usage.
src/java.desktop/macosx/native/libawt_lwawt/awt/AWTWindow.m line 1547:
> 1545: AWTWindow *awtWindow = [AWTWindow getTopmostWindowUnderMouse];
> 1546: if (awtWindow != nil) {
> 1547: topmostWindowUnderMouse = awtWindow.javaPlatformWindow;
I wonder what type of "reference" we should return here, I do not remember how
the "NewWeakGlobalRef" behave when returned to java.
src/java.desktop/macosx/native/libawt_lwawt/awt/CDragSourceContextPeer.m line
58:
> 56: jobject gTriggerEvent = (*env)->NewGlobalRef(env, jtrigger);
> 57: jlongArray gFormats = (*env)->NewGlobalRef(env, jformats);
> 58: jobject gFormatMap = (*env)->NewGlobalRef(env, jformatmap);
All above should be checked for NULL since OOM may occur, but it looks like it
does not throw OOM?
https://docs.oracle.com/javase/7/docs/technotes/guides/jni/spec/functions.html#NewGlobalRef
-------------
PR: https://git.openjdk.java.net/jdk/pull/2133
