[gcc 3.3.2] std::numeric_limits<long double>::singaling_NaN() bug?

[Christian Parpart]

I'm having a library that compiles under 3.4.4 and 4.0.1, however, when 
compiling under 3.3.2 (default compiler on SuSE Linux 9.3) I get the 
following error message

error: syntax error before `__builtin_nanf`

raised by the following fragment:

typedef long double float80;

class TFloat80 : ... {
...
    static const float80 NAN;
...
};

const float80 TFloat80::NAN = std::numeric_limits<float80>::signaling_NaN();

Why does it bail out here? I suppose this is a GCC/STL bug, however, is there 
a way to walk around this for GCC 3.3 builds?

Regards,
Christian Parpart.

-- 
 05:56:16 up 148 days, 19:03,  1 user,  load average: 0.45, 0.70, 1.02
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 189 bytes
Desc: not available
URL: <https://gcc.gnu.org/pipermail/gcc-help/attachments/20050819/437dabb8/attachment.sig>