On Tue, 27 Jan 2004 10:00:04 -0000, "n!" <nf********@nomailplease.com>
wrote:
Not sure why. You'd think it would be qualified by the 'using'
statement. Is this according to spec? I thought I had seen this done
without the error message.
This is down to the way the language determines what your declaration is
referring to. When looking up a typename, the language checks locally before
trying other areas of the code. As the namespace 'Somename' is within the
namespace you're working in (I guess you're at global scope, in the example)
it finds that first. If you are within an unrelated namespace, it should
fail to find 'Somename' and then proceed to look inside the 'Somename'
namespace (because of your 'using' statement). It will also work if your
code referencing the 'Somename' class is within the 'Somename' namespace, as
it will find the class first in its search :)
Hmmm, I think I made that as clear as mud, sorry.
Ha! About as clear as my illustration permitted, I guess. Yes, I
should have clarified that the caller is inside another namespace.
Revised:
file 1:
---------------------------------------------------
namespace Somename {
public class Somename {
....
}
}
file 2:
--------------------------------------------------
using Somename;
namespace Anothername {
class Anotherclass {
public static void Main() {
Somename s1 = new Somename; // Error!
Somename s2 = new Somename.Somename; // OK
}
}
}
--------------------------------------------------
Yes, I do see what's going on; the caller is looking for namespaces
first. It finds one, then (even though that namespace was already
called out in the 'using' statement) it assumes that it is looking at
a namespace ref and not a class ref. It looks no further.
I thought that I had seen code that implied the compiler handled this
more 'intelligently'. I guess not. For the number of silly things
that they *did* do (overloading 'new' and 'using' require analysis of
context), I thought they could handle it.
