On Wed, 2015-09-30 at 17:43 -0500, Michael Catanzaro wrote: > Minor issue: this and the next patch introduce the requirement that > accept is called on the data offer before the mouse grab is released. > I > now see this has always been required by mutter -- not sure why -- > but > it's new for Weston. (In fact, sending accepts has always been > optional.) The new restriction seems a bit arbitrary and could be > hard > for some clients. Chrome's cross-platform DnD abstraction all but > requires drag data to have been received before it can determine > whether to accept a mime type, so Chrome doesn't send an accept until > it's done receiving. Previously, that just meant the cursor would not > be updated in response to motion events during that time, but now it > means the drop will fail. This is not a huge deal, and the other > changes in these patches are much more important, but it suggests we > shouldn't add this restriction unless there's a good reason behind > it... and I don't know of one, so I'd like to understand that. The > change isn't necessary for the protocol as-is, and doesn't seem > necessary for DnD actions, either?
Thinking about this more, there is at least an obvious workaround: I can just send a bogus accept after receiving the drag_enter event (and then change it once the data is received.) _______________________________________________ wayland-devel mailing list wayland-devel@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/wayland-devel
