Remove some redundant function calls #463
Conversation
| @@ -1808,9 +1810,6 @@ outputmgrapplyortest(struct wlr_output_configuration_v1 *config, int test) | |||
| else | |||
| wlr_output_configuration_v1_send_failed(config); | |||
| wlr_output_configuration_v1_destroy(config); | |||
|
|
|||
| /* TODO: use a wrapper function? */ | |||
| updatemons(NULL, NULL); | |||
There was a problem hiding this comment.
This function is actually needed. While testing it turned out that without this call some changes didn't apply.
BTW, it can be replaced with wl_signal_emit_mutable(&output_layout->events.change, output_layout);
There was a problem hiding this comment.
Weird. Do you remember what kind of changes didn't apply, and are we sure it's not a wlroots bug? These lines came from cb01ce9, in case that helps jog your memory.
(I don't have an external on me right now, so I had to test using the wl backend instead of drm.)
dwl.c
Outdated
| @@ -1531,7 +1534,6 @@ void | |||
| maplayersurfacenotify(struct wl_listener *listener, void *data) | |||
| { | |||
| LayerSurface *l = wl_container_of(listener, l, map); | |||
| wlr_surface_send_enter(l->layer_surface->surface, l->mon->wlr_output); | |||
There was a problem hiding this comment.
As far I remember this is needed.
There was a problem hiding this comment.
Nevermind, it's not needed, but I found a funny thing, seems like the surfaces don't receive the wl_pointer.enter at all (with and without send wl_suface.enter ourselves)
There was a problem hiding this comment.
I discovered that swaynag is a better test (guessing dtao and bemenu are binding to older protocol versions), and there I'm still seeing both surface-enter-output and pointer-enter-surface:
$ WAYLAND_DEBUG=1 swaynag -m hello |& grep -Fw -e enter -e leave
[ 631375.417] wl_surface@10.enter(wl_output@8)
[ 633471.854] wl_pointer@9.enter(1155, wl_surface@10, 1476.41406250, 33.74218750)
[ 635047.065] wl_pointer@9.leave(1156, wl_surface@10)
[ 636230.590] wl_pointer@9.enter(1159, wl_surface@10, 1490.90625000, 32.12109375)
For the record, bemenu showed keyboard-enter with a similar test.
There was a problem hiding this comment.
Oh, I meant, it does not receive wl_pointer.enter until the cursor is moved, to reproduce:
- run
WAYLAND_DEBUG=1 fnott - move the cursor where the notification will be placed
- send a notify
EDIT: fnott link: https://codeberg.org/dnkl/fnott
sevz17
force-pushed
the
reduce-redundancy
branch
from
8d71123
to
c352729
Compare
|
Merged "No need to send surface.leave/enter events" and "Style: use early-return to clarify code" I still need to check the call of updatemons |
The output manager in wlroots emits an output_layout.change event when anything changes, so updatemons will be called anyway. ΔSLOC: -1
sevz17
force-pushed
the
reduce-redundancy
branch
from
c352729
to
b1cea77
Compare
Spotted a couple of things that wlroots already does for us, and a couple of places where we were called
printstatus()after making no changes.Net ΔSLOC: -5