[Bug c++/13963] gcc rejects array of unknown bounds as parameter of function template specialization

["austern at apple dot com" <gcc-bugzilla at gcc dot gnu dot org>]

------- Additional Comments From austern at apple dot com  2004-02-01 23:10 -------
Subject: Re:  gcc rejects array of unknown bounds as parameter of function template specialization

On Feb 1, 2004, at 11:58 AM, pinskia at gcc dot gnu dot org wrote:

>
> ------- Additional Comments From pinskia at gcc dot gnu dot org  
> 2004-02-01 19:58 -------
> The mangled symbol ICC produces is _Z1hIA0_iEvT_ which corresponds to 
> "void h<int [0]>(int [0])"
> which does not take your theory of "int[] should be treated as a 
> synonym for int*".

Mangled names (especially of templates) aren't a good guide to how the 
compiler interprets function types.  Remember that the mangled name of 
a template must keep track of the template argument exactly as written, 
because it needs to account for specializations and such: template 
parameters can be used for other things than just functino parameters, 
and the section of the standard that I quoted refers only to 
adjustments on function parameters.

Here's a better test: ask icc what it thinks is the type of the 
function template specialization h<int[]>.  If you apply typeid to 
h<int[]>, icc gives you "FvPiE".

And, of course, the test that began all this is that gcc thinks that 
h<int[]>(A) is illegal while icc accepts it.  Based on my reading of 
the standard (see above), I believe that icc is right and gcc is wrong. 
  Does anyone have an argument from the standard that says that gcc's 
behavior is right?



-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=13963