most of the time the activation token is passed
while the window is still unmapped. in this case
store the intend to activate the window for
later retrieval on map.
Okay, this might be one of the oldest layout issues to have remained uncaught.
Well, maybe as I add more randomized tests, I'll catch even more of those.
In sway, focus-follows-mouse keeps working during DnD, but not in niri.
So it can be surprising when you DnD something into another app, but it
doesn't get automatically focused. This commit fixes that.
Even if the DnD is not validated, or if there's no target surface (e.g.
dropped on the niri background), focus the target output, since that's
how Firefox's drag-tab-into-new-window works for example.
This way, expel becomes symmetric with consume. This is also how it
works in PaperWM. Though, in PaperWM if the expelled window was focused,
it will remain focused, while in this commit it is never focused, making
it the exact opposite of consume.
Use consume-or-expel-window-right for the old expel behavior.
* Unhide the pointer on scroll events
Since we reset the surface under the pointer when we hide the pointer
(see update_pointer_contents), scroll events don't work when the pointer
is hidden.
So to make scrolling work, we make sure that we unhide the pointer when
a scrolling event occurs.
* Update src/input/mod.rs
---------
Co-authored-by: Ivan Molodetskikh <yalterz@gmail.com>
This has the following benefits:
1. connector_connected() is now more closely mirroring
connector_disconnected() in that it merely lights up the connector,
and doesn't check if the connector should be off from the config.
2. We can use more complex on/off logic that depends on multiple
connectors. For example, this commit adds logic to only disable the
laptop panel on lid close if there are other connected outputs.
We don't want to disable the laptop panel on lid close if it's the only
connected output because it causes screen lockers to create their
surface from scratch on normal laptop unsuspend, which is undesirable
and also confuses some screen lockers.
valid tokens will stay around until explicitly cleaned-up.
remove the token after it has been successfully used
or we consider it timed out to prevent leaking the memory
used by the activation tokens
Ivan Molodetskikh