Bug191 Editing
Using the M1 release, there is a problem with the rendering of scrollbars (or sliders, whatever) when z rotation is used and the scrollbar thumb is dragged. First do the following: 1) Start the MVC demo. 2) Move the MVC window to the top left corner of the screen. 3) Open the settings dialog. 4) Zoom in as far as possible. 5) Z rotate until the MVC window is sitting on its right side (-90 degrees). 6) Close the settings dialog. Now move any of the 3 scrollbars. Note that only half of the scrollbar thumb actual gets rerendered, so it looks like the thumb gets cut in half (ouch!) lengthwise. Any change in the settings dialog forces a redraw that makes the thumbs look right, at least until they are dragged again. Another thing to notice is if you drag the thumbs in the left->middle->right (really bottom->middle->top) order so that all three thumbs are messed up, you'll notice that dragging any thumb correctly redraws any thumbs to the right of it. It should be noted that the movement of the MVC window to the top left of the screen zooming in were done merely to show off the effect better; they did not seem to have any effect on the bug. Also, if the window was instead rotated to about -91 degrees the problem didn't show up, but if it was rotated to -89 degrees the problem persisted. Also, if at -89 degrees, the thumbs were not cut in half, but instead cut with at a diagonal, so the line separating the redrawn portions from the non-redrawn portions is completely vertical (with respect to the screen). Also note that this problem occurs at other angles, but the angles given above were easiest for explaining the problem.
The wrong placement is coming directly out of Slider::extension (which in turn passes the ball to GraphicImpl::default_extension). The Region that comes in is indeed the correct size and shape, which means that the transformation must be bogus at this time. The real question is, why is it different from the matrix being used for drawing? The following is a snip from some debug output I'm taking: --ALLOCATE-- [ [ 6.12303e-17 -1 0 890] [ 1 6.12303e-17 0 1095.75] [ 0 0 1 0] [ 0 0 0 1] ] --DRAW-- [ [ 6.12303e-17 -1 0 810] [ 1 6.12303e-17 0 1095.75] [ 0 0 1 0] [ 0 0 0 1] ] Clearly, it's translated a bit too far to the side, and that's the kentyman is seeing. I still don't know which piece of code actually owns the bug, unfortunately.
I don't see this anymore. Did I fix this with my most recent commit?! Can somebody please confirm wether this bug is gone and close it if appropriate?
verified fixed.
Assigned to nicholas as he closed this one.
QcyAo3 <a href="http://kkisbraqzuim.com/">kkisbraqzuim</a>, [url=http://iqcaztsnfirl.com/]iqcaztsnfirl[/url], [link=http://dtisxujufoiq.com/]dtisxujufoiq[/link], http://bmqdasvgpnsi.com/
not logged in
register
Home