Re: Bug#513272: Availability of roundl() on armel?

["Jan C. Nordholz" <jckn@gmx.net>]

Hi Michael,

> I am trying to fix #513272, a FTBFS on armel. The problem basically
> boils down to the following snippet failing to compile on armel (tested
> with a Debian lenny on a qemu-emulated armel system):

see /usr/include/bits/mathdef.h on armel:

] [...]
] #ifndef __NO_LONG_DOUBLE_MATH
] /* Signal that we do not really have a `long double'.  This disables the
]    declaration of all the `long double' function variants.  */
] # define __NO_LONG_DOUBLE_MATH  1
] #endif

which has exactly the effect you describe. ;-)


Regards,

Jan