the white hole reveals text. when the window is resized the hole does not change size, until the page is refreshed. the white hole should resize as the window is changed by the user: graphics within a viewbox normally maintain scale as the window is resized. this is demonstrated in the attachment by a black circle which changes size as the window is resized by the user.
Created attachment 16170 [details] demo
I've created a simpler test (which doesn't use <use>) which I'll attach shortly.
Created attachment 16401 [details] simpler test case
My simpler test seems to pass in TOT, but I'm not sure if it's fully correct. The original "demo" fails. I think there was a %-sized <use> regression in TOT which is tracked by a separate bug. Yes, bug 14639.
Seems to work now?
5 Simon, unfortunately this is comprehensively broken.. on OS X please try the original demo... neither example scales any longer... the point being the mask has to scale compare opera/mozilla evidently if neither the object nor mask scale one cannot check this bug...
Using a current nightly build, I did not see obvious issues with either test case.
simon are you using os x? if so intel or ppc? what did you observe? ie have you compared results when resizing window with mozilla and opera? using os x with ppc and the latest nightly build: the simpler test does not resize the demo test mask is offset please confirm your observations using ppc or intel
The simpler test says "you should only ever see green", and I see just green. Nothing resizes when I resize the window. It's totally unclear what the 'demo' test should look like, but I do see one big black circle towards the left, which does not coincide with the white circle with text in.
Simon, please read the reduced test case code and understand a bug description before commenting.... the code would show you that the size is set as a percentage. this means it should scale or resize relative to the window. at the time the simpler test case was created it did, there is a regression bug, possibly filed, by me regarding this aspect. the simplest way to check whether there is a bug when unsure is as suggested in previous comments to check with a standards compliant browser such as opera or mozilla in this instance. naturally webkit is referred to in other bugs at mozilla and opera in some cases. Obviously in some instances it is necessary to take the difference to the working group for adjudication, but not in this case afaik.
Will be fixed with bug 43031. *** This bug has been marked as a duplicate of bug 43031 ***