Re: [orca-list] Writing Regression Test Question

["Brian G. Merrell" <bgmerrell novell com>]

On Fri, Sep 05, 2008 at 01:13:27PM -0400, Willie Walker wrote:
> Hi Brian:
>
> > I asked this question in IRC yesterday, but didn't get a response so
> > I am trying the mailing list.  So I apologize if someone is seeing
> > this question twice.
>
> Sorry about that.  The very very small Orca team has been heads down trying 
> to solve known issues for the GNOME 2.24 freezes coming up.  So, we have 
> been in hiding a little bit.

not a problem

> > I am writing some Orca regression tests to get familiar with Orca's
> > test harness so we can use it on our Mono WinForms accessibility
> > project (http://www.mono-project.com/Accessibility). 
>
> Cool!
>
> > Notice that the last second argument of the AssertPresentationAction
> > method has duplicate BRAILLE LINE and VISIBLE strings; this is the
> > only way the test will pass.  The strange part is that when running
> > Orca with orca.debug.LEVEL_INFO set, I don't see this duplication
> > when executing the key commands manually.  Any ideas why this would
> > happen?
>
> In looking at the Orca code, it looks to me that the lines are the result 
> of the automatic exiting of flat review when you cause something in the 
> interface to get focus.  That is, prior to pressing tab, flat review was 
> enabled and on the label above the buttons.  When you pressed tab, focus 
> was given to a button.  As it does with this and other events (e.g., caret 
> movement events), Orca automatically gets out of flat review mode and shows 
> the current object with focus.  It then continues to process the focus 
> event and shows the object with focus (again).

That helps a lot!  Thanks!

> We could potentially change this double-braille behavior in Orca so you 
> don't have to put multiple lines in your tests.  It's very close to the end 
> of the GNOME 2.24 cycle, however, and I'd rather leave the double-braille 
> behavior in there and remain conservative for now.  If you feel very 
> strongly about it, though, please file a bug in bugzilla and we'll track 
> it.

Now that I know that cause and know what to expect I am not too
concerned with it.  It does contradict the instructions on the
Writing Tests wiki, as instructs testers to use the
orca.debug.LEVEL_INFO output; if you consider it a bug I will be
happy to report it (not expecting it to be fixed for the next
release).

> Hope this helps!

Very much, thanks again!

Brian G. Merrell