Google Groups no longer supports new Usenet posts or subscriptions. Historical content remains viewable.
Dismiss

Placement of the using keyword

18 views
Skip to first unread message

Brian Gideon

unread,
Mar 11, 2005, 1:07:36 PM3/11/05
to
I stumbled across something odd today about the placement of the using
keyword. Section 9.3.2 of the C# v1.1 specification did not answer my
question. My confusion is isolated to what happens in File1.cs of the
following code. Notice that when the using keyword is placed outside
of the namespace decleration its behavior is different than placing it
inside the namespace decleration.

Questions:

1) Section 9.3.2 of the specification does not mention the differences
in the placement of the using keyword. Why? Is it described somewhere
else in the specification?

2) How does the compiler know to import A.B types as opposed to B types
when "using B" is contained within namespace A.C.D? Does the using
keyword somehow search the namespace hierarchy it is contained within
(in this case A.C.D) to find a match first and then search outside as a
last resort?

// File1.cs
// The following line imports namespace B types (Bar1 & Bar2).
using B;
namespace A.C.D
{
// The following line imports namespace A.B types (Foo1 & Foo2).
using B;
}

// File2.cs
namespace A.B
{
public class Foo1 { }
public class Foo2 { }
}

// File3.cs
namespace B
{
public class Bar1 { }
public class Bar2 { }
}

Thanks,
Brian

Sean Hederman

unread,
Mar 11, 2005, 2:29:06 PM3/11/05
to
"Brian Gideon" <brian...@yahoo.com> wrote in message
news:1110564456.1...@o13g2000cwo.googlegroups.com...

>I stumbled across something odd today about the placement of the using
> keyword. Section 9.3.2 of the C# v1.1 specification did not answer my
> question. My confusion is isolated to what happens in File1.cs of the
> following code. Notice that when the using keyword is placed outside
> of the namespace decleration its behavior is different than placing it
> inside the namespace decleration.
>
> Questions:
>
> 1) Section 9.3.2 of the specification does not mention the differences
> in the placement of the using keyword. Why? Is it described somewhere
> else in the specification?

Just had a look at Helsbergs book, and it appears that the only difference
in the placement relates to the scope of aliases:

// File1.cs
using R = N1.N2; // N1.N2 contains a class called OtherClass

namespace A {
class Class1 : R.OtherClass { // OK
}
}
namespace B {
class Class2 : R.OtherClass { // OK
}
}

// File2.cs
namespace A {
using R = N1.N2;
class Class1 : R.OtherClass { // OK
}
}
namespace B {
class Class2 : R.OtherClass { // Not OK
}
}

> 2) How does the compiler know to import A.B types as opposed to B types
> when "using B" is contained within namespace A.C.D? Does the using
> keyword somehow search the namespace hierarchy it is contained within
> (in this case A.C.D) to find a match first and then search outside as a
> last resort?

Interesting, from what I can see in the documentation, using B in A.C.D
should import B & A.C.D.B. If you're right (and I assume you are), it
probably means that it rolls up the namespace hierarchy: importing B & A.B &
A.C.B & A.C.D.B.

Brian Gideon

unread,
Mar 14, 2005, 9:42:37 AM3/14/05
to
Sean,

Thanks for the feedback. I also noticed that the differences in
placement of the using keyword were limited to normal scoping rules.
Section 9.3.1 is basically what you've cited.

I suspect you may be right about rolling up the namespace hierarchy.
But you'd think that would be mentioned somewhere in the specification.
If it isn't then we can only make assumptions based on observations.

Brian

Sean Hederman

unread,
Mar 14, 2005, 4:30:29 PM3/14/05
to
"Brian Gideon" <brian...@yahoo.com> wrote in message
news:1110811357....@f14g2000cwb.googlegroups.com...

> Sean,
>
> Thanks for the feedback. I also noticed that the differences in
> placement of the using keyword were limited to normal scoping rules.
> Section 9.3.1 is basically what you've cited.
>
> I suspect you may be right about rolling up the namespace hierarchy.
> But you'd think that would be mentioned somewhere in the specification.
> If it isn't then we can only make assumptions based on observations.

Probably is mentioned in subsection n, clause 18, paragraph 216 or
something. You'd probably have to do a full lexical analysis just to find
anything related ;D

0 new messages