Re: [orca-list] Please test master

[Joanmarie Diggs <jdiggs igalia com>]

Thanks for the update. I guess the next question is: Is this a recent
regression in Orca or something broken which you just now came across?

And if it's reliably reproducible a debug.out would be helpful.

Thanks again!
--joanie

On Wed, 2020-12-09 at 20:42 +0300, Alexander Epaneshnikov via orca-list
wrote:
> 09.12.2020 20:30, Joanmarie Diggs пишет:
> > And you're sure that's due to my change to use AT-SPI2's cache for
> > children?
> it isn't, sorry. next time I will check better.
> > On Wed, 2020-12-09 at 20:21 +0300, Alexander Epaneshnikov via orca-
> > list
> > wrote:
> > > 09.12.2020 18:01, Joanmarie Diggs пишет:
> > > > Hey all.
> > > >
> > > > I just removed a workaround in Orca master that potentially can
> > > > impact
> > > > all applications. We're early enough in the release cycle that
> > > > I
> > > > feel
> > > > comfortable making this change -- as long as we get lots of
> > > > testing.
> > > >
> > > > As for what changed:
> > > >
> > > > Until now, Orca was not using AT-SPI2's cache for accessible
> > > > children.
> > > > The reason why is because applications and toolkits have
> > > > historically
> > > > not been very good about emitting children-changed events when
> > > > objects
> > > > were added or removed. But not using this cache comes at a
> > > > cost:
> > > > performance. Those of you who noticed how much more performant
> > > > Orca
> > > > became with Chromium after I starting using the AT-SPI2 cache
> > > > for
> > > > accessible parents might be able to appreciate the difference.
> > > >
> > > > Unfortunately, using the cache *might* come at a cost: stale
> > > > trees
> > > > due
> > > > to applications/toolkits failing to emit those needed events.
> > > > That's
> > > > why we need testing: to catch those offenders. When we find an
> > > > offender, hopefully we can get them to fix their bug. If we
> > > > cannot,
> > > > we
> > > > can add a more targeted workaround so non-offending
> > > > apps/toolkits
> > > > can
> > > > be more performant.
> > > >
> > > > Anyhoo, please let me know what you find. Thanks!
> > > > --joanie
> > > hm i see a problem in mate-terminal.
> > > 1. run info: read top node.
> > > 2. press h for tutorial; for me text is garbled. lines too short
> > > or
> > > mismatched.