On Fri, Feb 23, 2007 at 06:23:34PM -0500, Blair Sadewitz wrote:
> In the "else if (dx || dy)" case (after the part that is patched),
> should dx*dx+dy*dy also instead be
> (float)dx*(float)dx+(float)dy*(float)dy?

That part is using floating point anyway. The former was meant to
workaround specific reoccuring bugs in the signal handling code of both
NetBSD and DragonFly. I'd avoid having to extend that unless someone can
show that the issues apply to other cases as well.

Joerg