Re: So much for "poor man's partial template specialization" :(

From: William Kempf (sirwillard_at_[hidden])
Date: 2000-10-26 14:02:25

--- In boost_at_[hidden], Jeremy Siek <jsiek_at_l...> wrote:
> Each template parameter must appear in a deduced context at least
> once in the argument list of a template function, and type 
expressions
> including :: are typically nondeduced contexts. So the test()
> function below is illegal.
> 
> 14.8.2.4
>   If a template parameter is used only in nondeduced contexts and 
is not
>   explicitly specified, template argument deduction fails. 

Ok, live and learn.  I'd say the error produced was ambiguous, 
however, though that's forgivable.  Some weird trivia, however:

Remove the call_traits stuff only from the second parameter and it 
compiles fine.  Similarly remove the second parameter all together 
and it compiles.  Remove the call_traits stuff only from the first 
parameter and the compiler still complains about an ambiguity.

Guess because I violated the standard I found some strange corner 
behavior on the compiler ;).

I guess I'm now a tad confused about the utility of call_traits.  I 
can see how it would help with template classes in some cases, but is 
there another way to do what was being attempted here with a template 
function?

What about the other problem I had?  Did I make another obscure 
error, or was the compiler at fault?

I've done a lot of reading about templates, including advanced 
techniques, so I thought I knew what I was doing here.  Guess the 
fact that I've never used advanced features because of the compiler 
I'm stuck with led me to at least one erroneous conclusion :).

Bill Kempf