volunteer needed to complete shadow header work

Ulrich Drepper drepper@redhat.com
Tue Aug 15 12:02:00 GMT 2000

Benjamin Kosnik <bkoz@redhat.com> writes:

> > > ../bits/std_limits.h: In function `long double std::numeric_limits<long 
> > > ddouble>::min ()':
> > > ../bits/std_limits.h:834: no field `__convert_long_double_i' in union 
> > > bbeing initialized

There is such a field.  I don't know what the C++ parser sees, though.

> [...]
> I've long wondered about this weirdness. Ulrich or Richard, do you know why 
> the current x86 code does this? It causes problems for c++ and long double 
> (this has also messed up the mknumeric_limits script in the past.)

Until recently there was no other way to ensure correct
initialization.  Now we have hexadecimal floating-point numbers.  But
does the C++ compiler understand them?

---------------.                          ,-.   1325 Chesapeake Terrace
Ulrich Drepper  \    ,-------------------'   \  Sunnyvale, CA 94089 USA
Red Hat          `--' drepper at redhat.com   `------------------------

More information about the Libstdc++ mailing list