std.traits and std.string incompatible ?

[Jonathan M Davis]

On Tuesday, March 08, 2011 13:24:44 Nick Sabalausky wrote:
> "Ali Çehreli" <acehreli at yahoo.com> wrote in message
> news:il5pge$nrr$1 at digitalmars.com...
> 
> > On 03/08/2011 08:24 AM, Wilfried Kirschenmann wrote:
> > > Hi,
> > > 
> > > When running the following file:
> > > 
> > > #!../dmd2/linux/bin/rdmd -unittest
> > > import std.string, std.traits;
> > > void main(string[] args){
> > > bool test = isNumeric(args[0]);
> > > }
> > > 
> > > I get the error :
> > > dmd2/linux/bin/../../src/phobos/std/traits.d(2576): Error: template
> > > std.traits.isNumeric(T) is not a function template
> > > 
> > > Is this a bug or is there something deprecated ?
> > > 
> > > Wilfried
> > 
> > isNumeric is a template. You are supposed to give it a type:
> >   if (isNumeric!SomeType)
> > 
> > or at compile time:
> >   static if (isNumeric!SomeType)
> > 
> > It doesn't work with string values. Although unnecessary, you could do
> > 
> > this:
> >     bool test = isNumeric!(typeof(args[0]));
> 
> No, there's an isNumeric in *both* std.traits and std.string. The one in
> std.traits is a template that takes a type. But the one in std.string is a
> function that takes a string and checks if the value of the string is
> numeric.
> 
> I'm on the latest D2 (2.052) and I just tried the example and got the same
> result. But if I *only* import std.string then it works. So it sounds like
> a bug: I forget the exact details of the rules involving overloading
> across modules, but one of two things should happen with the original
> example:
> 
> A. It should know that you meant std.string.isNumeric because of how you're
> calling it.
> 
> or:
> 
> B. It should complain that there's an ambiguity between
> std.string.isNumeric and std.traits.isNumeric and require you to
> disambiguate with either "std.traits." or "std.string."
> 
> I'm not sure which of those it's supposed to do, but it's clearly not doing
> either, so I'd file it as a bug: http://d.puremagic.com/issues/

isNumeric in std.string is a function. In std.traits, it's an eponymous 
template. The eponymous template should require the !. There's no function to 
feed the argument to. There shouldn't be any ambiguity of any kind. Overload set 
rules and whatnot should have nothing to do with this. This is definitely a bug.

- Jonathan M Davis