It's interesting to me that many Math functions do not specify any bound on precision, even though it's possible to have some bounds (like Java does) or to even specify the algorithm since free ones exist (fdlibm being the one I'm familiar with).
I am curious if this has been brought up before (quick and dirty search found nothing for me) and if there is some reason for not having a tighter spec, beyond lack of a champion.
Hi everyone!
It's interesting to me that many Math functions do not specify any bound on precision, even though it's possible to have some bounds (like Java does) or to even specify the algorithm since free ones exist (fdlibm being the one I'm familiar with).
I am curious if this has been brought up before (quick and dirty search found nothing for me) and if there is some reason for not having a tighter spec, beyond lack of a champion.
-Filip
It's interesting to me that many Math functions do not specify any bound on precision, even though it's possible to have some bounds (like Java does) or to even specify the algorithm since free ones exist (fdlibm being the one I'm familiar with).
I am curious if this has been brought up before (quick and dirty search found nothing for me) and if there is some reason for not having a tighter spec, beyond lack of a champion.