https://issues.apache.org/bugzilla/show_bug.cgi?id=43411
--- Comment #8 from Thomas Deweese <[email protected]> 2009-01-04 15:09:57 PST --- (In reply to comment #7) > tja - somehow you might be right. and i might collect the data and construct a > simplified case for their big bugbase. However: > - do we keep our bugs open for sun's ? > - i studied the values going into graphics2d: the glyphvector holds only > float-values, and the scale given previously just holds float-precision. so > when the path-elements are finally converted to their internal > coordinate-system to have the scan-lines probing the outlines, it doesnt have > enough 'resolution' to produce correct results. The fill of a shape should _never_ result is this rendering. A fill is always closed if it wasn't originally. So that single pixel high solid fill line is a bug in the coverage calculation in the rasterizer. Trust me I've worked on rasterizers and that failure is a clear indication of incorrect coverage calculation. > - it is an interesting mind-puzzle to imagine if there is a constallation, in > which the scale-value with just float-precision is the 'correct' value. This > might exist, but it is (i assume) far away from our test-case: we have scaling > in the font-size and in the canvas, and both influence the rasterizer. Generally neither the float or double is correct, the double may be closer but is rarely correct. Since the source content is a TrueType font in the case it only has ~11 bits of precision anyway (em square of 2048) a float value as ~25 bits of mantissa. > i would close this until someone comes up with another example. I'd rather back the commit out (as it involves a fairly large number of float casts "to no end") and close the bug saying it's not a bug in Batik but a bug in the JDK - if you really want to close the bug. All the commit has done is move the bug from a where we can reproduce it to someplace we can't reproduce it. If Archie wants to keep the patch because it helps him he is welcome to. -- Configure bugmail: https://issues.apache.org/bugzilla/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are the assignee for the bug. --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
