Re: [orca-list] Please test master in case I broke the universe

[Joanmarie Diggs <jdiggs igalia com>]

On Mon, 2021-06-14 at 01:03 +0300, Alexander Epaneshnikov wrote:
> ok. d261ce21e is the first bad commit. and judging by the
> description, this is highly expected.
> it seems to me that the orca does not handle the terminal focus event
> correctly.

You're correct about the commit, but not quite right about the
reason. Orca does handle the focus event correctly -- and/or figures
out focus when there's a missing focus event from the terminal. And it
activates the script for embedded terminals. So far so good.

But what was happening is that we're sometimes getting object:state-
changed:active on the window *after* we've correctly handled (or
figured out) focus. These late "hey this window is active!" events were
causing Orca to switch back to the script for the app. This was already
happening in the case of window:activate coming in late, which is why
some users reported this being an old problem. d261ce21e just made
things worse by also switching scripts in response to late state-
changed:active events on the window.

Both the regression and the long-standing issue are hopefully fixed in
master.

Thanks for the testing and the bisect!
--joanie