Hi Sergey,
1. That would be nice to test that modifiers are exactly as expected to
ensure that no extra modifier bits are set.
2. Test summary and author are omitted
3. In the test you are creating a new window for each button. That is
not optimal for test execution time.
4. Test fails on Linux, but you've filed a separate bug already. That
should be ok.
--Semyon
On 11/18/2015 6:50 PM, Sergey Bylokhov wrote:
Hello.
Please review the fix for jdk9.
On macosx when we create a KeyEvent we ignore the mouse state of all
mouse buttons, which means that mouse modifiers are missing.
The only KeyEvent is affected, because when we convert NS modifiers to
java modifiers we use nsToJavaKeyModifiers(), but in all other cases
we use nsToJavaMouseModifiers() which is the same but adds mouse
modifiers.
In the fix:
- The button parameter of nsToJavaMouseModifiers() was removed because
it was unused.
- nsToJavaMouseModifiers() was renamed to nsToJavaModifiers() and now
is used when keyEvent is generated.
- nsToJavaKeyModifiers() was removed because it unused now.
The similar bug for extended(id>=3) mouse buttons was filed for linux:
https://bugs.openjdk.java.net/browse/JDK-8143240
Bug: https://bugs.openjdk.java.net/browse/JDK-8143054
Webrev can be found at:
http://cr.openjdk.java.net/~serb/8143054/webrev.00