D fixed a lot of predecessor's mistakes :) Il giorno lun, 28/11/2011 alle 16.59 +0200, so ha scritto:
> On Mon, 28 Nov 2011 16:53:11 +0200, Andrea Fontana <[email protected]> > wrote: > > > dmd 2.056. > > > > void main(string[] args) > > { > > writeln("int.max: ", int.max); > > writeln("int.min: ", int.min); > > writeln("float.max: ", float.max); > > writeln("float.min: ", float.min); > > } > > > > it prints: > > int.max: 2147483647 <-- no int > int.max > > int.min: -2147483648 <-- no int < int.min > > float.max: 3.40282e+38 <-- no float > float.max > > float.min: 1.17549e-38 <-- this shoud be -float.max (or -inf?). It's > > not a min... > > > > assert(-1 < float.min); // passed! > > This drives me crazy on neural network... > > > > D doing exactly same thing what it's predecessor doing. > For example, everyone mostly define their own FLOAT_MIN, in C/C++.
