• Thorsten <thorstenkiefer gmx.de> Mar 03 2007
• "Jarrett Billingsley" <kb3ctd2 yahoo.com> Mar 03 2007
• Thorsten Kiefer <thorstenkiefer gmx.de> Mar 04 2007
• Sean Kelly <sean f4.ca> Mar 05 2007
• Thorsten Kiefer <thorstenkiefer gmx.de> Mar 03 2007
• Chris Nicholson-Sauls <ibisbasenji gmail.com> Mar 04 2007

Thorsten <thorstenkiefer gmx.de> writes:

Hi, when I compile this :

template toString(T) {
	char[] toString(T x){
		return "<??>";
	}
}
template toString(T : T[]) {
	char[] toString(T[] x){
		return "<array>";
	}
}

, I get this : 
tools.d(162): template tools.toString(T : T[]) specialization not allowed for
deduced parameter T

According to the online manual this should be no error.
How can I solve this ?

"Jarrett Billingsley" <kb3ctd2 yahoo.com> writes:

"Thorsten" <thorstenkiefer gmx.de> wrote in message 
news:esd2dn$126l$1 digitalmars.com...
 , I get this :
 tools.d(162): template tools.toString(T : T[]) specialization not allowed 
 for deduced parameter T

 According to the online manual this should be no error.
 How can I solve this ?


You are instantiating this template implicitly, as:

char[] s = toString([1, 2, 3]);

The spec says "Function template type parameters that are to be implicitly 
deduced may not have specializations." 

Thorsten Kiefer <thorstenkiefer gmx.de> writes:

Jarrett Billingsley Wrote:
 The spec says "Function template type parameters that are to be implicitly 
 deduced may not have specializations." 


And why is that ? The type parameters could be deduced and then the
best specialization could be used. Why do they exclude each other ?

Sean Kelly <sean f4.ca> writes:

Jarrett Billingsley wrote:
 "Thorsten" <thorstenkiefer gmx.de> wrote in message 
 news:esd2dn$126l$1 digitalmars.com...
 , I get this :
 tools.d(162): template tools.toString(T : T[]) specialization not allowed 
 for deduced parameter T

 According to the online manual this should be no error.
 How can I solve this ?


 You are instantiating this template implicitly, as:
 
 char[] s = toString([1, 2, 3]);
 
 The spec says "Function template type parameters that are to be implicitly 
 deduced may not have specializations." 


Really?  I'm almost sure I've done this before too.  How weird.


Sean

Thorsten Kiefer <thorstenkiefer gmx.de> writes:

OK, I solved it that way :

char[] toString(T)(T x){
	static if(is(T B == B[])) {
		char[] s;
		s ~= '[';
		if(x.length > 0){
			s ~= toString(x[0]);
			foreach(y;x[1..$]){
				s ~= ',';
				s ~= toString(y);
			}
		}
		s ~= ']';
		return s;
	} else {
		return std.string.toString(x);
	}
}

Chris Nicholson-Sauls <ibisbasenji gmail.com> writes:

Thorsten Kiefer wrote:
 OK, I solved it that way :
 
 char[] toString(T)(T x){
 	static if(is(T B == B[])) {
 		char[] s;
 		s ~= '[';
 		if(x.length > 0){
 			s ~= toString(x[0]);
 			foreach(y;x[1..$]){
 				s ~= ',';
 				s ~= toString(y);
 			}
 		}
 		s ~= ']';
 		return s;
 	} else {
 		return std.string.toString(x);
 	}
 }
 


You could also get this same effect just by calling std.string.format(), which
as I recall 
formats arrays in precisely this way.

format("%s", [1, 2, 3]) => "[1, 2, 3]"

(Admittedly its been a little while since I used Phobos, though.)

-- Chris Nicholson-Sauls