Re: [orca-list] FLAG WEEKEND (aka refactor part 2)

[Joanmarie Diggs <joanmarie diggs gmail com>]

Hi again, Steve.

I've got some good news and some bad news. Here's what I'm seeing using
Thunderbird 3.2a1pre from a few months ago:

* Flat review is working the same way for both the stable gnome-2-30
branch of Orca and for master. "Same" is less than ideal, namely:

* In a short list of messages, it seems we can't figure out what's
focused and what's not, so flat review is starting at the top of the
window. But flat review is otherwise working as I'd expect.

* In a huge list of messages (My 'All Mail' folder claims 27,273
messages, flat review takes forever to come up).

* The only tracebacks I'm seeing are intentional stacktraces we include
in debug.out to tell us that a node which should be alive has since
died/gone STATE_DEFUNCT. Gecko likes to kill it's accessibles for sport
and replace them with new versions. Kinda like Invasion of the Body
Snatchers, but with AtkObjects. But I digress... There are no tracebacks
suggestive of a refactor-related error. (Whereas when Jacob reported his
problem, there was indeed such a traceback indicating where I'd
forgotten to make a change.)

So I agree that in a really large list of messages, building the flat
review context is insanely slow for Thunderbird. And that's no good and
we should figure out where things are failing. But I'm not so sure it's
a refactor issue. Having said that, if you have a traceback for an Orca
error (not a dead-node stacktrace), see if it says that a given method
could not be found. If you find that, I blew it. And as soon as I know
where, I'll commit a fix.

Thanks for the testing and feedback!
--joanie