932254 died a horrible death, with a very ugly fix (adding the miterlimit to the bounding box in the nr-arena-shape update function).
My gut feeling is that the renderer must not have any preconditioned bounding box; it must _return_ one, not take. But this may be too difficult to implement. For now, adding the miterlimit length to the bbox sounds OK to me.
So the wrong cut problem is now gone, but the stroke-miterlimit value is still calculated wrong. According to spec, the value of sqrt(2) must correspond to the miter/bevel transition at the angle of 90 degrees, while now it happens at 45 degrees. We must conform to what the spec says to be compliant:
http://www.w3.org/TR/SVG11/painting.html#StrokeProperties
932251 i cannot produce either, but i had the exact same problem a while ago. it should have been fixed. if you can provide me a svg of the problem to debug...
I attached a sample svg to the bug.
_________________________________________________________________ http://join.msn.com/?pgmarket=en-ca&page=byoa/prem&xAPID=1994&DI...