3.1.1: hash_map don't work

3.1.1: hash_map don't work

[Igor Markov]

   Greetings,

   The following code gets compiled w/o problem by g++ 3.0.4,
   but doesn't with g++ 3.1.1

#include <ext/hash_map>
using std::hash_map;
main() {}

/n/gagarin/z/imarkov/devel/ /usr/bin/g++-3.0 test.cxx
/n/gagarin/z/imarkov/devel/ /usr/bin/g++-3.1 test.cxx
test.cxx:4: `hash_map' not declared
/n/gagarin/z/imarkov/devel/ uname -a
Linux gagarin.eecs.umich.edu 2.4.18 #6 SMP Mon Jul 15 12:00:46 EDT 2002 
i686 unknown
/n/gagarin/z/imarkov/devel/

   Can anyone reproduce this?
   If not, I should bother my sysadmin, who installed both 3.0 and 3.1.
   Otherwise, is it a bug or a feature?
   Should I expect 3.2 to behave the same way as 3.1 or as 3.0 ?

    thanks,

					  	    Igor
--
   Igor L. Markov         (734) 936-7829       EECS 2211
   http://www.eecs.umich.edu/~imarkov

Re: 3.1.1: hash_map don't work

[Andrew Pinski]

I can reproduce it but it is not a bug, but a feature.
hashmap has moved to a different namespace: __gnu_cxx.

Thanks,
Andrew Pinski


On Monday, August 19, 2002, at 09:22 , Igor Markov wrote:

Re: 3.1.1: hash_map don't work

[Igor Markov]

   Ok, thanks -- that worked.
   Is there a document with such gotchas?

   For gcc developers:
    If at all possible, it would be nice for g++
    to be more intelligent with error diagnostics
    in such cases. Having moved things around,
    it would make sense to add some code to intercept
    obvious error messages and add clarifications.

						 Igor
Andrew Pinski wrote:

Re: 3.1.1: hash_map don't work

[Phil Edwards]

On Mon, Aug 19, 2002 at 09:53:36PM -0400, Igor Markov wrote:
> 
>     Ok, thanks -- that worked.
>     Is there a document with such gotchas?

http://gcc.gnu.org/libstdc++/

The FAQ contains changes.  The extensions page lists things such as hash_map.
The source documentation lists the contents of namespaces.  All of these
show up in searches.


>     For gcc developers:
>      If at all possible, it would be nice for g++
>      to be more intelligent with error diagnostics
>      in such cases. Having moved things around,
>      it would make sense to add some code to intercept
>      obvious error messages and add clarifications.

It's difficult to do so correctly and consistently, and easy to get it wrong.
People are welcome to contribute such a thing, but the developers have limited
time and are not likely to write such wrappers.

Phil

-- 
I would therefore like to posit that computing's central challenge, viz. "How
not to make a mess of it," has /not/ been met.
                                                 - Edsger Dijkstra, 1930-2002