This is an automated email from the git hooks/post-receive script.
git pushed a commit to branch wl/browser-all
in repository enlightenment.
View the commit online.
commit 6ef50358a0aa893fd4e6453277a9dd62cec4389b
Author: Cedric BAIL <[email protected]>
AuthorDate: Mon Aug 10 20:17:31 2026 -0600
e_comp_wl - tell a client the pointer moved when the surface did
A surface that slides under a cursor that has not budged leaves the
client believing the pointer is still wherever it last was. The position
on the canvas really has not changed - the position within the surface
has, and that is the one wl_pointer.motion carries.
_e_comp_wl_pointer_focus_reeval() looks like it should already cover this:
it feeds evas a mouse move at the pointer's current position. It cannot.
Evas drops a mouse move whose position equals the previous one - the
cur_pt/prev_pt guard around the MOUSE_MOVE callback in evas_events.c - so
feeding the same coordinates delivers nothing at all, and the helper has
been a no-op for every caller that did not also move the pointer.
Nudging the pointer a pixel and back gets past the guard and does make the
tests pass, at the price of telling the client the pointer visited a place
it has never been. Send the one motion that is true instead, from the move
callback, in the surface's new frame of reference.
Interactive moves stay out of it: the pointer is driving those, and
move_end already re-evaluates once the grab is over.
wlcs: 640 passed -> 647, no regressions. Fixes
surface_moves_while_under_pointer and subsurface_moves_under_input_device
once/twice on wl_shell, xdg stable and xdg v6. The touch variants of the
same three tests still fail - E keeps no touch point state to recompute
from - and are handled separately.
---
src/bin/e_comp_wl.c | 35 +++++++++++++++++++++++++++++++++++
1 file changed, 35 insertions(+)
diff --git a/src/bin/e_comp_wl.c b/src/bin/e_comp_wl.c
index 65379544d..b9c2bea51 100644
--- a/src/bin/e_comp_wl.c
+++ b/src/bin/e_comp_wl.c
@@ -824,6 +824,30 @@ _e_comp_wl_evas_cb_move(void *data, Evas *e EINA_UNUSED, Evas_Object *obj EINA_U
ec->client.y + sec->comp_data->sub.data->position.y);
}
_e_comp_wl_surface_outputs_update(ec);
+
+ /* The surface moved under a cursor that never budged, so the pointer sits
+ * somewhere else in it now and the client has not been told. Nothing about
+ * the canvas position changed - the surface-local one did, and that is what
+ * wl_pointer.motion carries.
+ *
+ * Re-feeding evas at the unchanged canvas position is the obvious way to do
+ * this, and it does not work: evas drops a mouse move whose position equals
+ * the previous one, so no MOUSE_MOVE is ever delivered. Nudging the pointer
+ * a pixel and back does get past that, at the price of telling the client
+ * the pointer visited somewhere it has never been. Send the one motion that
+ * is true instead.
+ *
+ * An interactive move is the exception: the pointer is driving that one and
+ * move_end re-evaluates once the grab is over. */
+ if (ec->mouse.in && ec->comp_data->surface &&
+ (!e_client_action_get()) && (!e_client_util_ignored_get(ec)))
+ {
+ Evas_Coord px, py;
+
+ evas_pointer_canvas_xy_get(e_comp->evas, &px, &py);
+ _e_comp_wl_send_mouse_move(ec, px, py,
+ (unsigned int)lround(ecore_loop_time_get() * 1000));
+ }
}
static void
@@ -2956,11 +2980,22 @@ _e_comp_wl_client_cb_move_begin(void *data EINA_UNUSED, E_Client *ec EINA_UNUSED
static void
_e_comp_wl_pointer_focus_reeval(void)
{
+ static Eina_Bool reevaluating = EINA_FALSE;
Evas_Coord x, y;
if ((!e_comp) || (!e_comp->evas)) return;
+
+ /* Delivering the move can move something else - a subsurface following its
+ * parent, a client reacting to the enter it just got - and land us back
+ * here. One pass settles it; the nested ones would only ask the same
+ * question at the same coordinates. */
+ if (reevaluating) return;
+ reevaluating = EINA_TRUE;
+
evas_pointer_canvas_xy_get(e_comp->evas, &x, &y);
evas_event_feed_mouse_move(e_comp->evas, x, y, 0, NULL);
+
+ reevaluating = EINA_FALSE;
}
/* The other half of e_comp_wl_pointer_focus_drop(). xdg-shell hedges here -
--
To stop receiving notification emails like this one, please contact
the administrator of this repository.
