Created attachment 326204 [details] [review]
Patch for eel-canvas

Created attachment 326205 [details] [review]
Patch for canvas-container

Created attachment 326206 [details] [review]
Patch for dnd

Created attachment 326207 [details] [review]
Patch for tree-view-drag-dest

My limited testing revealed no regressions, but proceed with caution still.

Created attachment 326217 [details] [review]
Patch for eel-canvas

Messed up a call argument.

Review of attachment 326205 [details] [review]:

LGTM thanks!

Review of attachment 326206 [details] [review]:

ditto

Review of attachment 326207 [details] [review]:

sure!

Review of attachment 326217 [details] [review]:

Timestamp parameter is no longer necessary, so you can remove it.
Also, I don't have the knowledge to review it actually :/ sorry for that. Specifically:

::: eel/eel-canvas.c
@@ +873,3 @@
+				  NULL,
+				  NULL,
+				  NULL);

I'm not sure this is equivalent. Where is provided the event_mask? (I checked and I have no idea neither).
We need to check with gtk+ developers. Would you like to approach them in IRC on channel gtk+ on irc.gnome.org?
garnacho or mclasen are probably the people to reach.

Pushing the ones that are ok

Created attachment 326798 [details] [review]
Patch for eel-canvas

gdk_seat_grab() does not take an event mask, because it grabs all the events, corresponding to the capabilities. In this case, all pointing capabilities (recommended, unless we /really/ only want the pointer only). The event type is checked against the mask in emit_event() if the item is grabbed, so, AFAICT, we’re in the clear.

I eliminated the NULL arguments (at a cost) in this one. Now, instead of time, eel_canvas_item_grab() takes an event argument, which it passes to gdk_seat_grab(), so all the calling functions have been modified accordingly. It’s still fine to pass NULL, however. The second NULL argument before was the prepare_func. While not completely necessary (in this case (I think)), it prepares the window for grabbing (in this case (also in the majority of cases), makes it visible).

Also touched up the documentation, because some comments were outdated.

Changed the status of pushed patches.

Review of attachment 326798 [details] [review]:

Yeah this makes sense to me now, thanks!

Attachment 326798 [details] committed as 225f2cf - eel-canvas: use GdkSeat operations

Oops, sorry about that.

Attachment 326798 [details] pushed as 225f2cf - eel-canvas: use GdkSeat operations