GNOME Bugzilla – Bug 472301
Consume behavior not conforming to IDL
Last modified: 2019-03-27 20:12:13 UTC
Hi, Currently pyatspi has an DeviceEvent attribute called 'consume' that could be set to true to stop the trickling of the event to other observers and ultimately the desktop. This attribute is not part of the IDL. Peter, was this attribute created to solve a certain problem? Perhaps in LSR? If not I would recommend removing it and allowing observers to return True to consume the event.
Created attachment 94720 [details] [review] Proposed patch This patch enables event callbacks to return True to consume the event. For uniformity I also got rid of Event.consume and put the same logic in place.
Created attachment 94945 [details] [review] Proposed patch This patch does not deprecate the consume attribute. This is more practical to commit now because of the proximity to a GNOME release.
(In reply to comment #2) > Created an attachment (id=94945) [edit] > Proposed patch > > This patch does not deprecate the consume attribute. This is more practical to > commit now because of the proximity to a GNOME release. > I think this looks OK. Li, how about you? It would be nice to get this into the GNOME 2.20 release.
I'm sorry, I must missed the mail. Is it worth to break hard code freeze? Or we can wait until 2.20.1
I think it is worth putting in, the earlier the better so that dependent projects don't resort to using the consume attribute. This will cause breakage in later releases.
The patch doesn't break any API/ABI and string freeze, right?
Nope. It doesn't.
OK. I think we can try to break the freeze. Could you send mail to release-team to explain the reason we want to break the freeze?