Arthur, my apologies if I was wrong: I had my message posted to this forum given that I was not looking for help (I had one workaround implemented and I can think of another workaround which is to simply ignore the received event) but rather wanted to share my findings. I was hesitant to directly submit a bug because other OSS comunities (like RedHat) require users to describe bugs in a forum first...
--- Sami, my understanding is that the suggested fix (with the else clause) is only being called whenever a user calls UIObject.sinkEvents () or UIObject.unsinkEvents()... which is not performed very often, is it? I mean, it is not called during event propagation. So why worry about the throughput issue? (BTW, 99% is a slightly exagerated figure for the fall-back into the 'else' clause: there are <30 events defined in the release I use (GWT 1.5.2) and in many cases, the bit Event.ONMOUSEWHEEL might be actually set). Finally looking at the W3C spec for these add/remove functions (see link below) it is harmless to add a listener twice (i.e. it won't be called twice) as well as removing a non-registered listener. Not sure how well Mozilla implements that specification though... http://www.w3.org/TR/DOM-Level-2-Events/events.html#Events-Registration-interfaces So I will now go down the route to submit a bug and... see what happens next. Thanks for your attention so far. Yann --~--~---------~--~----~------------~-------~--~----~ http://groups.google.com/group/Google-Web-Toolkit-Contributors -~----------~----~----~----~------~----~------~--~---