> introduces the possibility of one ULP error in the calling arg. Not > big, admittedly, but try doing sind(180) vs. sin(pi) and see what the > results are. >
Slightly off topic, but I just wanted to note that are also sinpi and cospifunctions (although no other related functions) which give more accurate results for multiples of pi. julia> sin(pi) 1.2246467991473532e-16 julia> sinpi(1) 0 Cheers, Kevin > In any event, keeping them around as convenience functions is useful > for folks coming from (or porting code from) at Matlab background. > > Just my opinion. > > Stuart > > > > On Tue, 4 Feb 2014, Johan Sigfrids wrote: > > I agree with this. It seems like a lot of extra namespace usage. Besides, >> if degrees2radians() is too long you always could define a constant and >> use >> that: >> >> const DEGREE = pi/180 >> sin(90DEGREE) >> >> >> >> On Tuesday, February 4, 2014 6:57:16 PM UTC+2, Hans W Borchers wrote: >> >>> >>> I was astonished to see the following functions in the Julia Standard >>> Library, >>> functions that accept (or return) degrees instead of radians: >>> >>> sind asind secd asecd >>> cosd acosd cscd acscd >>> tand atand cotd acotd >>> >>> I didn't find these function names in any other technical computing >>> system. >>> Each of these functions can easily be reconstructed by the user applying >>> the >>> functions degrees2radians() and radians2degrees(). >>> >>> I feel these function names clutter the namespace of Julia Base without >>> being >>> of any real value. Therefore, I would like to vote for deprecating use of >>> these >>> functions in the next version of Julia. >>> >>> Hans Werner >>> >>> >>>
