itfi limitation or bug on my part?

[Brad Roberts]

On 8/7/2010 12:42 PM, Don wrote:
> Brad Roberts wrote:
>> module test;
>>
>> void main()
>> {
>>     int[] foos;
>>     func1(foos);
>> }
>>
>> void func1(T)(const T[] foos)
>> {
>>     T afoo;
>>     func2(foos, afoo);
>> }
>>
>> void func2(T)(const T[] foos, out T afoo)
>> {
>> }
>>
>> $ dmd -c test.d
>> test.d(12): Error: template test.func2(T) does not match any function template
>> declaration
>> test.d(12): Error: template test.func2(T) cannot deduce template function from
>> argument types !()(const(int[]),int)
>> test.d(6): Error: template instance test.func1!(int) error instantiating
>>
>> Ignore style.. this is a massively reduced case from a much more complex block
>> of code.
>>
> 
> Looks like a compiler bug. If you change func2 to this:
> 
> void func2(T, U)(const T[] foos, out U afoo)
> {
>    static assert(is(T==U));
> }
> 
> test0.d(212): Error: static assert  (is(const(int) == int)) is false
> test0.d(207):        instantiated from here: func2!(const(int),int)
> test0.d(201):        instantiated from here: func1!(int)
> 
> T should be "const(int)" instead of "int".

Other way around.. T should be 'int' instead of 'const(int)'.  Thanks for the
confirmation.  I thought the same thing but wanted independent confirmation.
I'll file the bug.