On Tue, Aug 28, 2007 at 08:59:05AM +0800, zou lunkai wrote: > > - float cx = x + in->read_sint(num_bits); > > - float cy = y + in->read_sint(num_bits); > > - float ax = cx + in->read_sint(num_bits); > > - float ay = cy + in->read_sint(num_bits); > > + int cx = x + in->read_sint(num_bits); > > + int cy = y + in->read_sint(num_bits); > > + int ax = cx + in->read_sint(num_bits); > > + int ay = cy + in->read_sint(num_bits); > ... > > Good to see these changes. It really meaningless to convert to float > numbers at parse time. And AGG seldom uses float numbers(double and > integral numbers are choosed internally, IIRC).
Unfortunately, edges are still made of floats, so conversion will still happen when creating them. I'm not sure about effects of making edges made of ints as they'll eventually be scaled and need to be floats anyway. We should probably make edge a template itself (and move under libgeometry). --strk; _______________________________________________ Gnash-commit mailing list Gnash-commit@gnu.org http://lists.gnu.org/mailman/listinfo/gnash-commit
