FWIW, that code is unlikely to conform to the C standard, in the sense that it is (probably, I think, almost certainly, well, pretty sure) undefined. You can't do casts like that, and so the authors are giving the C compiler the liberty to do whatever it wants.
Robby On Fri, Jul 12, 2013 at 11:02 AM, John Clements <cleme...@brinckerhoff.org>wrote: > > On Jul 12, 2013, at 7:17 AM, Greg Hendershott wrote: > > > > [1] http://blog.quenta.org/2012/09/0x5f3759df.html > > ^this. > > I was just skimming this blog post until I realized that they were doing a > bit shift on the IEEE floating-point representation. Madness! > > Many thanks. > > John > > > ____________________ > Racket Users list: > http://lists.racket-lang.org/users >
____________________ Racket Users list: http://lists.racket-lang.org/users