Hi Manaji,
Okay, back to werev.01.
Could you remove the comment in lines 262-268 since you assume that it
is not true anymore and so CGEventCreateKeyboardEvent/CGEventPost will
not cause any troubles.
Did you analyze why werev.02 fix did not pass those tests?
--Semyon
On 6/1/2016 4:46 PM, Manajit Halder wrote:
Hi Semyon and Sergey,
After running the tests shared by Sergey I found that the second
webrev (webrev.01) shared by me solves the problem.
http://cr.openjdk.java.net/~mhalder/8155740/webrev.01/
<http://cr.openjdk.java.net/%7Emhalder/8155740/webrev.01/>
Following tests were present in
https://java.net/jira/browse/MACOSX_PORT-621:
closed/java/awt/KeyboardFocusmanager/ConsumeNextMnemonicKeyTypedTest/ConsumeNextMnemonicKeyTypedTest
java/awt/Dialog/NestedDialogs/Modal/NestedModalDialogTest.java
java/awt/Dialog/NestedDialogs/Modeless/NestedModelessDialogTest.java
But only first test (which is now present as part of open code) could
be performed and the remaining tests were not found in the present code.
The first test fails due to another issue (JDK-8156460), whose fix is
in progress and will be send for after this issue. This fix
(JDK-8155740) is not related to the failure of the first test case.
The new location of the first test:
java/awt/KeyboardFocusmanager/ConsumeNextMnemonicKeyTypedTest/ConsumeNextMnemonicKeyTypedTest
Please review the webrev.01.
Thanks,
Manajit
On 26-May-2016, at 1:05 pm, Semyon Sadetsky
<semyon.sadet...@oracle.com <mailto:semyon.sadet...@oracle.com>> wrote:
On 5/24/2016 2:09 PM, Manajit Halder wrote:
Hi Semyon,
It is not clear in the comment what was the problem, but it looks
like the problem was observed just by
using CGEventCreateKeyboardEvent/CGEventPost. In my case I don’t see
any issues just by using the fix. It posts all the key events and
all the key events are tested in the test file modified along with
the fix.
If you found the comment is not actual anymore, why did you preserve it?
--Semyon
Apart from the unknown problem mentioned in the existing comment
this fix has following advantages:
* Key codes for modifier key are required to distinguish between
Alt and Alt-Gr key, which is not possible by using existing
solution as it doesn’t post key codes for modifier keys. And
also keys can’t be distinguished base on modifier value, which
is same for both keys (Alt and Alt-Gr).
* As mentioned in the existing
comment CGEventCreateKeyboardEvent/CGEventPost is a better solution.
* Online search about keyboard simulation showed
that CGEventCreateKeyboardEvent/CGEventPost is used to simulate
key stores in most of the cases.
* Tested the key codes, didn’t observe any problem.
Regarding number keys not working
with CGEventCreateKeyboardEvent/CGEventPost:
Solved the problem by providing event source
to CGEventCreateKeyboardEvent. Code is modified.
Please review the modified code:
http://cr.openjdk.java.net/~mhalder/8155740/webrev.02/
Thanks,
Manajit
On 20-May-2016, at 12:02 am, Semyon Sadetsky
<semyon.sadet...@oracle.com <mailto:semyon.sadet...@oracle.com>> wrote:
Hi Manajit,
Before your fix all key codes wa sent using
AXUIElementCreateSystemWide();
and CGEventCreateKeyboardEvent was commented or excluded from
compilation:
274 #if 0
275 CGEventRef event = CGEventCreateKeyboardEvent(NULL,
keyCode, keyPressed);
276 if (event != NULL) {
277 CGEventPost(kCGSessionEventTap, event);
278 CFRelease(event);
279 }
280 #endif
I just wonder why it was done. Maybe that was some other issue fix.
The comment above states:
262 * Well, using CGEventCreateKeyboardEvent/CGEventPost
would have been
263 * a better solution, however, it gives me all kinds of
trouble and I have
264 * no idea how to solve them without inserting delays
between simulated
265 * events. So, I've ended up disabling it and opted for
another approach
266 * that uses Accessibility API instead.
I don't understand what trouble is mentioned in the comment above.
But in your fix you've put the CGEventCreateKeyboardEvent back, may
it return this trouble back?
Also as I understand Numpad number keys did not work as well. Could
you add the corresponding test case since you provide a fix this
extra issue?
--Semyon
On 5/13/2016 11:50 PM, Manajit Halder wrote:
Hi Semyon,
The fix is changed a bit because it was observed that the modifier
keys plus alphabet keys were not working together. In the modified
fix only Num keys are posted by AXUIElementPostKeyboardEvent and
remaining keys are posted by CGPostKeyboardEvent/CGEventPost. The
fix is explained in the comment in file CRobot.m.
Please review the new changes:
_http://cr.openjdk.java.net/~mhalder/8155740/webrev.01/_
_
_
Answers to your questions:
What is difference between AXUIElementPostKeyboardEvent and
CGEventCreateKeyboardEvent?
Difference as per the documentation:
AXUIElementPostKeyboardEvent is similar to CGPostKeyboardEvent
(which synthesizes a low-level keyboard event on the local
machine), but it allows you to specify the target application as
opposed to always sending the events to the active application. If
the system-wide accessibility object is passed in the application
parameter, the event is sent to the active application.
Difference behaviour as per the implementation observed while
debugging the code:
AXUIElementPostKeyboardEvent:
AXUIElementPostKeyboardEvent posts 0 key code for all the modifier
keys with key codes 16, 17,18, 20, 157 and also for newly added
modifier key VK_ALT_GRAPH. But it posts correct key code for all
the remaining keys.
While debugging it was that for modifier keys keyDown and keyUp
events are not generated, but flagsChanged event (flagsChanged:
(NSEvent *)event) is generated. But for all other keys keyDown
followed by keyUp events are generated.
CGEventCreateKeyboardEvent:
CGEventCreateKeyboardEvent posts correct key code for all the keys
except for numeric keys (numbers 0 to 9 on normal
keyboard). CGEventCreateKeyboardEvent posts wrong key codes for
the number keys 0 to 9. Instead of posting number key codes it
posts Numpad key codes for the corresponding number key. For
example Numpad0 key code is posted for number 0, Numpad1 key code
is posted for number 1 and simillarly for remaining num keys.
Why the latter was commented? Does it mean that valid modifier
keys have not been sent by AWT robot?
I didn’t get your question clearly. If you meant why in the
current implementation the later part (fix using
CGPostKeyboardEvent) of fix was commented.
I am not very sure about it. As per the comment it is only clear
that CGPostKeyboardEvent/CGEventPost would have been a better
solution and I agree with that, perhaps reason could be related to
the difference in behaviour observed while debugging the code as
mentioned above.
Thanks,
Manajit
Hi Manajit,
Just a few questions to clarify on the fix.
What is difference between AXUIElementPostKeyboardEvent and
CGEventCreateKeyboardEvent?
Why the latter was commented? Does it mean that valid modifier
keys have not been sent by AWT robot?
--Semyon
On 5/12/2016 10:45 AM, Manajit Halder wrote:
Hi All,
Kindly review the fix for JDK9.
*Bug*:
https://bugs.openjdk.java.net/browse/JDK-8155740
_
_
*Webrev*:
http://cr.openjdk.java.net/~mhalder/8155740/webrev.00/
<http://cr.openjdk.java.net/%7Emhalder/8155740/webrev.00/>
*Issue: *
[macosx] robot.keyPress and robot.keyRelease do not generate key
event for Alt-Graph key VK_ALT_GRAPH.
*Cause: *
VK_ALT_GRAPH is a new key added to the Mac OS X platform and it
is not mapped to any key on the OS X platform.
*Fix: *
VK_ALT_GRAPH is mapped to right option (OSX_RightOption) key on
Mac OS X.
Method Java_sun_lwawt_macosx_CRobot_keyEvent is modified for the
following reason:
AXUIElementPostKeyboardEvent posts 0 key code for all the
modifier keys with key codes (16, 17,18, 20, 157) and also for
newly added modifier key VK_ALT_GRAPH.
But it posts correct key code for all the other keys. On the
other hand CGEventCreateKeyboardEvent posts correct key code for
all the modifier keys and
hence it is used to post modifier key events and
AXUIElementPostKeyboardEvent is used to post all the remaining
key events.
Regards,
Manajit
