On Sun, 2010-10-10 at 13:38 +0200, Karl Hammar wrote: > > I'm also very surprised that we are still discussion this. > > I got into this thread due to some faulty assumptions about floats.
I think most of us know, that floating point operation today is similar in speed than integer operations -- for modern INTEL/AMD x86 type processors. Of course sin/cos/sqrt is slower for integers, because it involves conversion int to float and back. It may be different for ARM/MIPS/POWER-PC/ATOM and others. Indeed using only floats may be the fastest for x86 architectures, because divide and sin/cos/sqrt... are native available. But as Armin noted, the not well working == is a problem, it needs very carefully handling. And having an exact grid, so that all coordinates are multiples of one base unit is very convenient. For a decision one has to investigate the existing code and planned additions. Problem is not trivial, I thing cairo graphics library is currently using fixed points integers for internal representation, but they do think about converting to floats. Luckily current PCB developers are smart, we have pointed them to all possible problems, so all should be fine. Best regards Stefan Salewski _______________________________________________ geda-user mailing list geda-user@moria.seul.org http://www.seul.org/cgi-bin/mailman/listinfo/geda-user