Ambiguity with class-virt-specifiers

[Inconnu]

What is defined in the declaration #1, according to the new standard:
new empty final class N::X or variable N::final of type ::X?

struct X {int x.y; };

namespace N
{
struct X final {}; // #1
}

 

[Balog Pal]

What is defined in the declaration #1, according to the new standard:
new empty final class N::X or variable N::final of type ::X?

struct X {int x.y; };

namespace N
{
struct X final {}; // #1
}

2.11p2
<quote>
The identifiers in Table 3 have a special meaning when appearing in a
certain context. When referred to in the grammar, these identifiers are used
explicitly rather than using the identifier grammar production. any
ambiguity as to whether a given identifier has a special meaning is resolved
to interpret the token as a regular identifier.
Table 3 - Identifiers with special meaning
override final
</quote>

According to that the ambiguity is resolved for the latter.

 

[Inconnu]

2.11p2
<quote>
The identifiers in Table 3 have a special meaning when appearing in a
certain context. When referred to in the grammar, these identifiers are used
explicitly rather than using the identifier grammar production. any
ambiguity as to whether a given identifier has a special meaning is resolved
to interpret the token as a regular identifier.
    Table 3 - Identifiers with special meaning
         override final
</quote>

According to that the ambiguity is resolved for the latter.

OK. Then I have 2 questions more. 1) Isn't it too hard a task for a
compiler to resolve such an ambiguity? Just imagine what it must do.
2) How to define empty final class X in the namespace N?

 

[Balog Pal]

"Inconnu" <[email protected]>

(forwarding from clc++ to csc++ as a better fit group)

OK. Then I have 2 questions more. 1) Isn't it too hard a task for a
compiler to resolve such an ambiguity? Just imagine what it must do.
2) How to define empty final class X in the namespace N?

That is a good question. I add another one. Let's just have
struct S final {};

That supposed to be definiton of S, right?
Now let's insert before that
struct S;

Does that change the interpretation?

For these cases disambiguation can be done toward object by removing the
class-key. I guess 2.11p2
was formed that way to guard the old code and not change meaning if someone
used the new contextual keywords as identifiers. But this seem to create
counter-intuitive thing like with the 'this really is a function
declaration' madness.

For this situation was the intent to use stuff between {} as decision-maker?
And force tricks like

struct S final {public:};
?

If so, I'd suggest at least put in a note about that at the proper place in
the standard text.

 

[Johannes Schaub (litb)]

What is defined in the declaration #1, according to the new standard:
new empty final class N::X or variable N::final of type ::X?

struct X {int x.y; };

namespace N
{
struct X final {}; // #1
}

Maybe you want to add yourself to the cc-list here:
http://llvm.org/bugs/show_bug.cgi?id=9922