Mail Archives: djgpp/1999/09/24/18:12:27
Hans-Bernhard Broeker wrote:
> C9x is still a moving target, and no efforts in the DJGPP community
> have yet been made to implement any of its changes. Quite a large part
> of it will have to be done by the gcc developers, not in DJGPP,
> anyway. The standardized set of library functions hasn't changed all
> that much, in comparison to the language itself.
Actually, the math functions will change quite a lot; each function will come in
three flavors: float, double, and long double. I expect that the "float" versions
could easily be derived from the "double" functions currently implemented, but the
"long double" functions will require a lot of new work. There will also be new
functions, including cbrt, erf, erfc, lgamma, tgamma, fmin, fmax, etc.
Furthermore, most of the transcendental functions will have complex versions,
though the draft standard currently says that an implementation is not required to
support complex arithmetic. gcc already supports complex arithmetic in a
nonstandard way, so that capability will almost certainly be standardized and
retained.
-Eric Rudd
rudd AT cyberoptics DOT com
- Raw text -