Fisheye rollover effect doesn't work properly when browser window maximised from "restored" state

[Spectoral]

--When the browser loads in "restored" window mode, fisheye rollover works.

--When the browser loads in "maximised" window mode, fisheye rollover works.

--When the browser is maximised from an initially "restored" mode, the fisheye still responds when the mouse rolls over the area of the screen that was occupied by the fisheye *prior* to the window being maximised. However, the fisheye will not respond with the rollover feature if you rollover the new location where the icons are. It should be noted though, that mouseclicks on the actual fisheye images still work as links, regardless of where the fisheye has been moved.

--When the browser is "restored" down from an initially maximised mode, the fisheye responds when the mouse rolls over the area of the screen that was occupied by the fisheye *prior* to the window being "restored". BUT; only if this initial area where the fisheye was, is still within the browser window's new 'smaller' dimensions. If the fisheye (in the browser's maximised state) sat outside the outskirts of the browser window once the browser is put into it's "restored" state, the fisheye won't respond at all to mouse rollovers. However, mouseclicks on the actual fisheye still work as links, regardless of where the fisheye has been moved.

--When the browser loads in "restored" window mode, and the window is click-n'-dragged to different positions on the screen, the fisheye works as it should.

--When the browser is "minimised" and then "restored" the fisheye works as it should.

--When the browser is "minimised" and then "maximised" the fisheye works as it should.

[tk]

This will be fixed in 0.9 SVN tonight...

until then, here is the fix action

//add this at the end of startup() to accomodate resizing
this._onResizeHandle = dojo.connect(window, 'onresize', this, '_initializePositioning');

//add this in the destroy routine with other disconnects
dojo.disconnect(this._onResizeHandle);

If Dante gets my email in time, he'll commit the fix, if not, I'll check it in later tonight.

-Karl

[dante]

(In [9869]) patch from Karl fixes #1519 and adds addition isFixed param to recalculate grid coords after page scroll (on fixed/absolutely positioned fisheyes) ... refs #637 (and should probably mark as wontfix 'fully' as it would be cumbersome to know when the fisheye is a child of a movable pane. making a public re-init function to call would fix fully (to connect to getParent().resize())?