• Tomek =?UTF-8?B?U293acWEc2tp?= (5/5) Oct 16 2010 Funny thing:
• dsimcha (9/12) Oct 16 2010 IMHO yes, this is a bug. What you wrote is syntactic sugar for:
• Rainer Deyke (5/9) Oct 16 2010 I think 'void foo(t)' is syntactically correct. You can't know that 't'

Tomek =?UTF-8?B?U293acWEc2tp?= <just ask.me> writes:

Funny thing:

void foo(T)(t); // typeless argument

If you don't instantiate the template, it actually compiles. Bug?

-- 
Tomek

dsimcha <dsimcha yahoo.com> writes:

== Quote from Tomek Sowiński (just ask.me)'s article
 Funny thing:
 void foo(T)(t); // typeless argument
 If you don't instantiate the template, it actually compiles. Bug?

IMHO yes, this is a bug.  What you wrote is syntactic sugar for:

template foo(T) {
    void foo(t);
}

The rule with uninstantiated template bodies is that the code needs to be
syntactically correct, but not necessarily semantically correct (since the
semantics can only be fully analyzed on instantiation).  void foo(t) looks
syntactically incorrect.

Rainer Deyke <rainerd eldwood.com> writes:

On 10/16/2010 19:26, dsimcha wrote:
 The rule with uninstantiated template bodies is that the code needs to be
 syntactically correct, but not necessarily semantically correct (since the
 semantics can only be fully analyzed on instantiation).  void foo(t) looks
 syntactically incorrect.

I think 'void foo(t)' is syntactically correct.  You can't know that 't'
isn't the name of a type without semantic analysis.


-- 
Rainer Deyke - rainerd eldwood.com