Paul.Zimmermann at inria.fr
Tue May 19 09:04:55 UTC 2015
> Well, you use mpn_com_n (it have never been in the official API), you
> #include "gmp-impl.h" ...
you are right. However if I want to only use the official API then:
* I have to use malloc instead of TMP_ALLOC_LIMBS (fine)
* I need invert_limb (referenced in the 6.0.0 manual but not in the API)
* I need mpn_add_nc (not in the official API)
> but you are right we should try to expose in the
> official API all the functions that can have a stable interface and can
> help programmers using GMP.
> mpn_zero_p is in gmp-impl.h since 2008:
> Do you feel it might be useful?
> We should probably add some more instruments to inspect and compare numbers.
> This loosely relates to a message I sent two years ago about mpz_cmp_ui
> (https://gmplib.org/list-archives/gmp-devel/2013-February/002807.html ).
> The questions are: do we prefer macros or (possibly inlined) functions for
> this tasks? Which instruments do we need?
I have no preference for macros or (possibly inlined) functions, but such
functions would be useful indeed.
More information about the gmp-devel