[C++-sig] [OT] gcc bugzilla [was: gcc 3.4 pre-compiled header support?]

Niall Douglas s_sourceforge at nedprod.com
Tue Jan 13 23:38:40 CET 2004


-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

On 13 Jan 2004 at 4:41, Ralf W. Grosse-Kunstleve wrote:

> --- Raoul Gough <RaoulGough at yahoo.co.uk> wrote:
> > In my experience, gcc bug reports lie dormant for an unpredictable
> > amount of time. Sometimes you get an analysis almost immediately
> > (especially if the report is simple and in error) and sometimes
> > nobody seems to notice it for six months or so.
> 
> I filed a few bug reports recently which were attended to within a few
> hours. The most important one is fixed already. I guess I got lucky,
> but I am posting this here to encourage people to submit bug reports.
> Chances are that it will help eventually.

Thanks to both of you for the advice. If it's normal then I rest 
easy.

> > The best way to get action on
> > it is to fix it yourself, of course. Some other suggestions that
> > might help - reduce the size of the demonstration code to the
> > absolute minimum (ten lines would be good) and test it using the
> > latest compiler sources, since it might already be fixed.
> 
> Niall wrote that all existing gcc's are broken. In a case like that
> I'd try very hard to come up with an alternative approach. It is
> always a big advantage if existing/native compilers can be used. In
> Niall's case I'd try to write a wrapper around std::vector<> that
> zeros out the memory in the destructor.

Well as I noted MSVC7.1 I think is broken here too - it shouldn't 
treat operator new differently to malloc(). However I can work around 
MSVC's misbehaviour whereas GCC's is a tad harder :(

I'll wait for v3.4 to get more stable, then try the test again and if 
it fails again, I'll submit another bug report with the same thing. 
It's not fantastically urgent for me as it only faults at process 
exit so I can test my code. Besides, from now on there's no platform 
specific code so theroretically what works on Windows works on Linux 
(famous last words!).

Cheers,
Niall





-----BEGIN PGP SIGNATURE-----
Version: idw's PGP-Frontend 4.9.6.1 / 9-2003 + PGP 8.0.2

iQA/AwUBQARzcMEcvDLFGKbPEQJVtACfdbYvqu4WGR63qZQbr4XdcZu5BIQAnRvc
/wP/Qn3jtIqSzcYsq0n9wiuj
=IO3I
-----END PGP SIGNATURE-----




More information about the Cplusplus-sig mailing list