Overloading external functions for use in a separate module and... uniform function call syntax!

[Atash]

Some of you StackOverflow-folks may have already seen this 
question. That said...

http://stackoverflow.com/questions/18320610/overloading-external-function-for-use-in-a-separate-module

That describes the question. Now... Andrei Alexandrescu was 
awesome enough to throw me a bone, but... no dice. Under the 
assumption that the double-spacing between the `runrun` decldef 
and the `import mod_a` represents a separation of files, GDC 
4.8.1 spits out "template mod.run does not match any function 
template declaration. Candidates are:" <the `run` decldef that 
doesn't accept doubles/>.

Given the feel of the situation, I wanted to check with 
dlang.org-folk to make sure that this isn't a hole I've 
inadvertently fallen into as far as D features go with respect to 
uniform function call syntax and the semantics surrounding it.

For the sake of discussion, I'm modifying my question a little 
bit... Let's say I have some function in some module:

= BEGIN CODE =

== file mod.d

module mod;
void doStuff(T)(T v) { v.doOtherStuff(); }

== file adooblydoo.d

module adooblydoo;
import mod;
struct A;
doOtherStuff(A a) { ... }
void main() { A a; doStuff(a); }

= END CODE =

This fails to compile with both DMD 2.063 and GDC 4.8.1 informing 
me that doOtherStuff is an undefined identifier while 
instantiating doStuff.

Given that Andrei Alexandrescu gave me an answer implying that at 
least the C++-STL-style of calling was possible, I'm inclined to 
guess that this is either a bugfeature or at the very least an 
obscure feature with respect to what I'd assume to be the 
intended semantics of uniform function call syntax.

Soooooo... What is this behavior?

(( THANKS FOR READING THAT WHOLE THING [if you didn't and you're 
reading this little message, I take the thanks back and instead 
chuck silly putty at your face] ))