[Bug c++/13358] New: long long and C++ do not mix well in 3.3/3.4

["lloyd at randombit dot net" <gcc-bugzilla at gcc dot gnu dot org>]

I would request that the warning 'integer constant is too large for "%s"
type' be controllable, preferably by -Wno-long-long. As of now (3.4 20031126,
and 3.3 last time I looked), using a long long constant in C++ source causes g++ 
to freak out. This following code works fine with egcs 1.1.2, gcc 3.0.4, 3.1,
3.2.2, and half a dozen commercial compilers

int main()
   {
   unsigned long long x = 0xFEDCBA9876543210;
   return 0;
   }

but g++ 3.[34] will barf on it, and the only way to get it to compile at all is
with -fpermissive (I would prefer a way to kill this warning completely). If
you wondering why I don't just add ULL to everything: VC++ and some other
pre-C99 Windows compilers don't like that (at all).

I suppose technically this warning might be triggered by something other than a
long long constant, making -Wlong-long the wrong flag to control it, but odds
are pretty damn good that a contant that won't fit in a long is a long long
constant, not a typo.

Anyway, it's a highly trivial fix in c-lex.c at line 556, just add
warn_long_long to the set of conditions. Since it's on by default for C/C++, 
this will not affect anyone except people who pass -Wno-long-long to the 
compiler. 

(If not -Wno-long-long, at least make some way to disable it besides
"-fpermissive -w", which is a really awful 'solution').

This actually effects C as well, but due to it being only a warning there, I'm
marking it as a C++ problem.

-- 
           Summary: long long and C++ do not mix well in 3.3/3.4
           Product: gcc
           Version: 3.4
            Status: UNCONFIRMED
          Severity: normal
          Priority: P2
         Component: c++
        AssignedTo: unassigned at gcc dot gnu dot org
        ReportedBy: lloyd at randombit dot net
                CC: gcc-bugs at gcc dot gnu dot org
  GCC host triplet: i386-pc-linux-gnu


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=13358