Hey, question.. Just wondering if gcc is supposed to behave this way now.
For example, I have a template:
template <class T> class foo
{
...
void method1(T a1) {...}
T method2() {...}
...
};
Notice, method1 accepts a parameter of type T and method2 returns a
value of type T.
Well, if foo is instantiated as foo<int> then everything is fine.
However, lets say we typedef an array type and instantiate foo with it...
typedef int bar[10]; // bar is a type: int[10]
foo<bar> f;
Now, method1 is accepting an array parameter (fine), but method2 is
returning an array value (error)
In gcc-3, there was no error until you tried to call f.method2(), but in
gcc-4 it gives the error when foo<bar> is instantiated whether the
method is called or not. I used this feature intentionally in gcc-3,
but I wanted to know if it's an ANSI thing one way or another. And if
it's not ANSI compliant now, then I guess it needs to be fixed in gcc.
Thanks,
Davy
