Re: [orca-list] Strange speech problem

[Joanmarie Diggs <jdiggs igalia com>]

Hi Samuel.

At the moment, I cannot think of why we'd want that. It seems rather
late in the release cycle to propose that be changed. But perhaps giving
it a try early in the next release cycle?

In the meantime, perhaps Hypra could dog food the proposed change before
any patch is submitted to the Gtk+ developers. If y'all have thoroughly
beat on the proposed change for a non-trivial amount of time without
negative side effects, THEN it's probably safe to do a merge request for
Gtk+. :)

Thanks!
--joanie

On 2/15/19 7:32 AM, Samuel Thibault wrote:
> Samuel Thibault, le ven. 15 févr. 2019 12:57:59 +0100, a ecrit:
> > Ah, running this again in my normal desktop session, I get the same
> > issue, I'll dig a bit.
>
> This seems to be due to this inside gtk
> (atk_key_event_from_gdk_event_key):
>
>   if (key->string && key->string[0] &&
>       (key->state & GDK_CONTROL_MASK ||
>        g_unichar_isgraph (g_utf8_get_char (key->string))))
>     event->string = key->string;
>   else
>     event->string = gdk_keyval_name (key->keyval);
>
> I.e. it assumes that when control is used, the litteral control
> character should be passed instead of the name of the key.
>
> I don't think we ever want this, since orca brings that back to the key
> name by hand already? The git log doesn't provide any rationale for
> this, at least.
>
> Samuel