diff options
author | Marius Vlad <marius.vlad@collabora.com> | 2020-06-27 18:20:19 +0300 |
---|---|---|
committer | Marius Vlad <marius.vlad@collabora.com> | 2020-07-08 20:40:39 +0300 |
commit | 3b0e548c3adae12e72cc6d6dc22f5ecc475aa7c2 (patch) | |
tree | 060896dd1f8a4484caf267afe0adf539d68864cd /app/afbclient.cpp | |
parent | 40e6e78a19483dc5822ded81cea041f3983a877f (diff) |
app: Re-work the deactivation part in application_id_state eventkoi_10.93.0koi_10.92.0koi_10.91.0koi/10.93.0koi/10.92.0koi/10.91.0jellyfish_9.99.4jellyfish_9.99.3jellyfish_9.99.2jellyfish/9.99.4jellyfish/9.99.3jellyfish/9.99.29.99.49.99.39.99.210.93.010.92.010.91.0
In commit 11b9762f115778 we've added some logic in place that would
allow to start the application de-activated (that is hidden), as to
avoid 'stealing' the input focus (hoovering the pointer over a surface
will give that surface focus).
alexa-viewer application would be started by default by the system and
will be set 'invisible' (due to being a Qt app with the property set as
visible: false), and most likely this might cause a bit
of confusion for users. Also, by default, the compositor will display
the application as soon as started.
This patch re-works the de-activation part to handle a corner case, as
in commit 11b9762f115778 we would send the de-activate request as a
response in an event handler but it seems we can't guarantee that at that
time the surface was indeed activated.
So, instead of using application_id event, we use the application_id_state
event, which guarantees that we were actually sending the de-activate
request after we got the activate state event from the compositor.
Bug-AGL: SPEC-3412
Signed-off-by: Marius Vlad <marius.vlad@collabora.com>
Change-Id: Ie50b8011700d258a899374176fb78dcd528f88c1
Diffstat (limited to 'app/afbclient.cpp')
0 files changed, 0 insertions, 0 deletions