Re: structure alignment problem ?
> > Right. Well, I'm not sure I'd class it as a bug: glibc just doesn't provide
> > long double math functions for ARM at the moment. And, in turn, the reason
> > for this is that GCC doesn't actually emit extended precision operations:
> > "long double" is equivalent to "double".
That's fine, except that the 'fabsl' and a couple of others long double functions
_are_ builtin the c++ compiler (mapped to double precision?). Unfortunatly the configure script is using fabsl
to detect long double support. i've changed the test to acosl for my package.
don't know how this autoconf stuff is generated, but i hope it does not put the
'fabsl' test in all its scripts around the world...
That's weird, why a few functions are correctly mapped and others not ?
BTW: info found in /usr/include/g++-v3/arm-linux/bits/c++config.h on debussy
> Does ANSI C99 say that it provides the long double maths functions?
> [even if sizeof(long double) == sizeof(double)]
> If yes, and if glibc is trying to give a full C99 implementation, then it's
> a glibc failing (or "bug"), isn't it?
> [even if all they'd are aliases to the double precision versions]
after some research:
C99 _does_ provides long double. all the math long double functions _are_ defined is
the white paper.
but, glibc _is not_ fully C99 compliant (according to their web page at gnu, a few
things of C99 are yet to be done / can't be done) . given that, it seems
glibc provides long double math on some architectures which support it
and not on others, seems to be defined partially in the file above.
Thanks for the pointers ;)
and again, sorry for the delay, i must be more careful about my responsabilities.
the package is twutils.