Red Hat Bugzilla – Bug 806065
Pressing Esc key in jBPM Console throws exception
Last modified: 2012-04-25 11:58:47 EDT
Created attachment 572073 [details]
Description of problem:
Esc key is often used to abort/cancel application dialogs. I'm often tempted to close the confirmation message after starting a process this way. There are many other dialogs the user might try to close using keybord. Hitting Esc always results in exception printed in the Messages panel.
Version-Release number of selected component (if applicable):
Steps to Reproduce:
1. log in into jBPM Console
2. press Esc key
Exception is thrown.
No exception should be thrown. If there is an active dialog, it should be closed.
see attached screenshot
I am trying to de-obfuscate the stacktrace, but without much success. Apparently the classes used in that stacktrace aren't in bpm-console itself:
2012-04-24 14:58:26,576 [ERROR] Communication Error<br/>Additional details:<br/> None
ClientMessageBusImpl sounds like Errai.
Yes, it's Errai 1.1.Final.
Problem is De-obfuscated that stacktrace doesn't help much: Errai 1.1's ClientMessageBusImpl effectively eats the server-side exception.
What I need is the server side log, which will have the real stacktrace.
Porcelli and me finally got bpm-console and jbpm-gwt-console to build from source and run.
However, some things to notice:
- there is no server side log of this exception. So either:
-- the server eats the exception
-- the client uses the wrong url.
Problem is that ClientMessageBusImpl even eats the status code.
The bpm-console doesn't support GWT hosted mode (not as part of jbpm-console anyway), so I 've hacked the errai jars to get some more info what goes wrong:
2012-04-25 16:01:22,747 [ERROR] Communication Error<br/>Additional details:<br/> None
ge0ffrey hack statusCode (0) text ()
Unfortunately that states the impossible: the HTTP statusCode is 0, which by definition always 1xx, 2xx, 3xx, 4xx or 5xx.
The text is empty.
Since the exception stacktrace is a dead end because there's no way to figure out what code activated the request or even what the request was, I 've looked at it from the other direction:
- There is only one usage of addKey*() in bpm-console: org.jboss.bpm.console.client.LoginView#createForm(). That code cannot cause this as it does nothing on an ESCAPE key down or pressed event.
- All other code that involves the word "key" isn't related to keyboard events.
This leads me to believe, it might be something in the libraries that we use.