if someone double clicks in url bar (and any content but url bar is most noticable) if they twitch the mouse on the second down click it starts a drag. This then snaps the selection. The bug usually reported is that people double click and the selection starts at beginning of word (or url) and stops at the mouse pointer instead of continuing to the end of the url.

adding keyword

this adds an api on nsIFrameSelection which remembers the mouse event point of the doubleclick. When the drag happens we use the api to get the old event point and check the distance. if the distance isnt great enough then bail out of drag. This has the affect of fixing our url problem. 5 pixels (75 twips) is the buffer zone to start dragging again. once you start dragging the

adding cc for saari

adding akkana cc list

akkana could I get a review on this patch? I will need an sr from kin as well. what do i have to do to get this on the radar to check in? adding kin to cc list.

forgot an interface header

The second patch seems to improve things quite a bit! I can still get the old evil behavior if I try to move the mouse between clicks, but I have to try. But can I ask what the old behavior is there for anyway? Does anyone really want or expect a click-while-moving to cause a selection from the beginning of line/word to the cursor position? Is there other software that behaves this way? I would have expected that any two clicks within the doubleclick timeout (I'm not sure where that's set -- you'd think it would be an nsILookAndFeel parameter, but it doesn't seem to be) would be considered a doubleclick, and select the current word. This new code is a lot better, but I'm not clear why we need the select-from-beginning-to-point code or the mouse position remembering code at all.

Re my comment on doubleclick time: looks like that's handled in the event code for each platform, so we'd just be looking at clickCount on the nsMouseEvent.

ok this one doesnt allow dragging after a double click until you leave the current selection.

ok caught my own error there in setmousedoubledown. This works and looks good.

I'm seeing the select-to-current-point behavior without moving the mouse at all. I'm using a stationary trackball with a big ball, so I'm sure it's not moving at all. Is this the same issue, or is that a new bug that needs to be filed?

I also saw it when I was quite sure my mouse hadn't moved, but Mike's patch seems to cure it for me. I'm guessing that something in the hardware (quantum fluctuations? :-) was reporting small movements even though the physical device wasn't moving.

marking this as dup of 23784 I will put patch up there. *** This bug has been marked as a duplicate of 23784 ***