Re: STL iterator problem

[jlh <jlh@xxxxxx>]

> gcc is right to reject this code, something more
> standard compliant should pass :
>
> for(typename B::iterator i; ...)

Yep, conveniently, gcc-3.4.3 even suggests it:

a.cpp:29: error: dependent-name ` B::iterator' is parsed as a non-type,
          but instantiation yields a type
a.cpp:29: note: say `typename  B::iterator' if a type is meant

But still I don't get why this is wrong.  Is only because the compiler
can't tell whether it's a typename or something else at parse time?

btw, take care with "parent<std::list<int>>".  Apparently GCC now
gets it right, but to be portable, better write "parent<std::list<int> >",
to avoid parsing a shift operator instead.

Cheers,
jlh

Attachment: signature.asc
Description: OpenPGP digital signature