After an evaluation, GNOME has moved from Bugzilla to GitLab. Learn more about GitLab.
No new issues can be reported in GNOME Bugzilla anymore.
To report an issue in a GNOME project, go to GNOME GitLab.
Do not go to GNOME Gitlab for: Bluefish, Doxygen, GnuCash, GStreamer, java-gnome, LDTP, NetworkManager, Tomboy.
Bug 105963 - Arrow keys are being ignored in Java Applications
Arrow keys are being ignored in Java Applications
Status: RESOLVED INCOMPLETE
Product: gnopernicus
Classification: Deprecated
Component: keyboard&mouse
unspecified
Other Linux
: Normal blocker
: ---
Assigned To: ps
ps
Depends on: 106081
Blocks: 105962
 
 
Reported: 2003-02-13 09:30 UTC by Darren Kenny
Modified: 2004-12-22 21:47 UTC
See Also:
GNOME target: ---
GNOME version: 2.1/2.2



Description Darren Kenny 2003-02-13 09:30:48 UTC
For some reason arrow key movements with Java Applications are somehow
being consumed by gnopernicus. This is making java apps quite unusable.
Comment 1 ps 2003-02-13 10:00:49 UTC
Hi.
I need more information about, in which application, on which OS
(Solaris v linux), and how can reproduce this bug.
Comment 2 Darren Kenny 2003-02-13 11:14:12 UTC
Specifically, I'm using Java's JFC demo program StylePad. I'm using
Linux RedHat 7.1 with the GNOME 2.0 nightly from Feb 2, 2003. (And
a slightly more recent build of gnopernicus).

To reproduce, run StylePad using a JDK with GNOME accessibility enabled:

  java -jar $JAVA_HOME/demo/jfc/StylePad/StylePad.jar

Run Gnopernicus.

  Now focus the StylePad window and type in text, then try to use
  the arrow keys (not the Num Pad arrows, but the seperate keys
  beside the NumPad) to move the caret around the text. This fails
  to do anything.


Comment 3 ps 2003-03-05 15:45:05 UTC
I made a workaround to does not eat the not interested keys, but the
issue is in the java-access-bridge because it is return a not register
keyevents. The modificatioin on CVS head.
Comment 4 Darren Kenny 2003-03-11 15:56:33 UTC
Didn't mean to close this one, yet... Sorry..
Comment 5 bill.haneman 2003-05-14 16:56:58 UTC
note: these are desired for key echo.
Comment 6 padraig.obriain 2003-05-21 08:52:43 UTC
Paul,

Are you saying that there is a bug in the Java Access Bridge? If so,
can you give more detail as I do not understand it.
Comment 7 bill.haneman 2003-05-21 10:30:38 UTC
Padraig: I think the problem is that somehow gnopernicus is consuming
the 'regular' arrow keys when java apps are running.  gnopernicus
needs to consume the numberpad keys, but not the 'normal' arrow keys.
 I suspect that the problem may be that for some reason gnopernicus'
request for numpad arrow keys is resulting in notifications for
(non-KP) arrow keys as well, resulting in those keys being consumed.

Assuming this is so, there are several potential causes and several
possible approaches.  Next step I guess is to identify exactly which
events are affected and why they are reported&consumed, and whether
there's a way to distinguish between them in Java.  If there is, then
j-a-b needs code to do this filtering I guess; or possibly there is a
j-a-b bug at work.  If there is not, then we may be blocked on Java.

Comment 8 ps 2003-05-21 12:26:30 UTC
Hi Bill:   
The arrows keys did NOT consumed by the gnopernicus. Gnopernicus   
after he get a event, it return the key event for other application.   
Gnopernicus consume only the numpad keys and the user defined keys.  
  
Padraig:  
I checked it again this bug whit a version of gnome 2.2 stack from  
09_05_2003.  
This bug I can not reproduce now. I think it fixed in at-spi or in 
j-a-b in the time. 
I describe what is the bug: At gnopernicus can regester key 
combinations, called user keys. (ex CTRL-a). Gnopernicus register 
this combinations.The registartion are made with keystring 
registartion. On gnome toolkit the keylistener listened only the 
registered keys, but on java toolkit was occure key event which is 
not regitered, in our example the arrows keys. In this listener all 
keys are consumed. And in this case the arrows keys was comsumed. 
I thinks the problem was with a registartion of keystrings, because 
at-spi returned "_"  keystring in java toolkit for arrows keys, 
perhaps one of the registration was made incorrect in server from 
at-spi/j-a-b, and the at-spi/j-a-b are registered whit this "_" 
string. After the client listened all keys which has a returned 
keystring "_". 
 
I say again. I checked it again, and I don't see this bug with this 
version of stack. 
Comment 9 bill.haneman 2003-05-21 12:45:35 UTC
when Paul mentions "_" keystrings, what he was really seeing were
ASCII/UNICODE control codes which are properly returned in the
"string" field for control-key combinations.

It may be that JABG returned control-strings in the "string" field for
arrow keys.  I think this should be investigated once more so that we
understand what was happenning here.

Paul, your description of why events were being consumed (i.e.
returned to the "wrong" listener) matches with what I was trying to
say earlier, when I said "gnopernicus is consuming the arrow keys";
gnopernicus was possibly consuming them because they were being
returned to a listener which had registered for some OTHER keys, I did
not mean that the error was in gnopernicus.

Comment 10 padraig.obriain 2003-05-21 14:55:58 UTC
I have checked the name reported when arrow keys are pressed in a Java
application. The names are "Up" "Down" "Left" and "Right". 
Comment 11 padraig.obriain 2003-06-11 09:57:00 UTC
As this bug cannot be reproduced with gnome 2.2 stack from 09_05_2003
I am closing it.