Tracking bug - to document Dogtail<->Frysk interaction problem on text fields
dynamically generated by GUI
This problem was first identified with Frysk as built from CVS head on 20060628,
but it's probably been there for a while.
The problem is that when Dogtail writes text into text fields that are
dynamically generated in the GUI - for example, by pressing the "+" button in
the custom observer dialog to create multiple instances of filter/action values
- Frysk does not recognize that text.
This problem is illustrated with test script
If you watch the script operate, you can see text being written into the fields.
When Frysk saves the custom observer, however, the text is not written to the
I have not been able to recreate the problem manually - only with Dogtail.
Interestingly, when I inserted some sys.sleep() statements into the script, and
manually appended a character to the string fields after Dogtail had written the
data into the fields, Frysk was able to recognize the text and write it to the
Anyway - this is causing problems for the automated tests - I'll pursue this
with the Dogtail folks and drag in Frysk people as needed...
Whatever this bug was - it's fixed now - with Frysk and Dogtail as built from
CVS on July 15, 2006.
Hang on - I was able to create fork type observers this morning and not
encounter the problem. But, this afternoon, it's failing with exec type observers.
Created attachment 1170 [details]
Data file to recreate test
Using this test data file recreates the failure with the
TestCreateObserversfromDataModel.py test script.
The problem occurs in trying to create observers of types: exec, task terminaction
The problem is not seen with fork, syscall, and task clone observers
These data files were just checked in: