[PATCH] Fix initialization of _M_grouping

[Paolo Carlini]

Paolo Carlini wrote:

> Ok if it passes?

Testing ok.

I have two questions:
1- What if the user does'nt have the it_IT tables installed? It is ok to fail in
that case? I think so, we have already tests which assumes the de_DE data, right?
2- Perhaps I had better to remove from the testcase the line

VERIFY( th == '\0' );

since this is specific to glibc and probably very soon, as indicated by Nathan, we
will adjust NUL to ',' as the thousands separator in the case of glibc, after
having inferred from it (as indicated by Ulrich) grouping == "", right?

Cheers,
Paolo.