Steven Schveighoffer wrote: > On Mon, 12 Apr 2010 13:45:14 -0400, Jérôme M. Berger <jeber...@free.fr> > wrote: > >> We are talking about range propagation, a function of the compiler, >> not a function of the compiled program. Since we can't get a 100% >> accurate representation of the possible values anyway (even yours >> might leave holes in the middle after all), then it might be better >> to choose a faster, slightly less precise algorithm if the >> difference is not too great. That's the difference between a >> compiler and a full-fledged static code analysis an program prover. > > When we're talking about the difference between O(1) and O(lgn), I'll > take accuracy over speed in my compiler any day. The solution should be > 100% accurate for the problem statement. If the problem statement is > not what we need, then we need a new problem statement :) Solving the > problem statement for 99% of values is not good enough. > And when we're talking about the difference between 10s and 55s for a minimal loss of accuracy, which will you take? Especially if the accuracy loss is less than is lost elsewhere (due to holes in the ranges).
>> >> Anyway, the point is moot, I have a new version of my algorithm >> with 100% precision and high speed. > > Yes, I'm still trying to understand how it works :) > :) Jerome -- mailto:jeber...@free.fr http://jeberger.free.fr Jabber: jeber...@jabber.fr
signature.asc
Description: OpenPGP digital signature