This is the mail archive of the libstdc++@gcc.gnu.org mailing list for the libstdc++ project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: Where are std::sinf( float ) etc.


Gabriel Dos Reis wrote:

>I would think that the special functions offer more interesting
>challenges :-)
>  
>
Indeed :-)

Then, my plan would be leaving to Ed that "interesting" task, and
instead, concentrate on Chapter 8, which still included facilities
requested by quite a few users.

By the way, something I noticed again, while looking into the details of
acos & co, is that, when, _GLIBCXX_USE_C99_COMPLEX is not defined, we
are providing texbook implementations, which likely don't fully conform
to the corresponding C99 specifications: think nans, infs, etc... I
guess we can keep on following this strategy, but now, for TR1, the
issue is much more evident, because Chapter 8 explicitely specifies the
functions in terms of C99, which, in turn, compared to C90, goes into
the details of nans, infs, etc (the famous Annex G)...

Well, I would say there isn't much point in providing very sophisticated
implementations when _GLIBCXX_USE_C99_COMPLEX is not defined, meaning
the underlying libc doesn't provide *at all* the corresponding C99
functions (cacos, etc.), and, well, I would suppose most implementations
are in the process of being upgraded to the C99 specifications anyway.

Paolo.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]