[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: Missing functions tanPi, asinPi and acosPi



Dear all,



crlibm contains correctly rounded implementations of asinpi, acospi.
atanpi should be feasible since the worst-cases are known.
They should be reincluded into the draft.

I agree with this. And yet, I have voted "approve" on the draft, because I don't
consider adding these three functions is something *very* important, it would
just be *reasonable*. After that, we will have to stop at some point: we cannot put a new function in the standard each time we manage to implement it correctly rounded. What was important was to standardize the most common functions (i.e.,
the basic trigs, the exponentials and logarithms in radices e, 2 and 10, and a
few other functions).

For nastier functions, what matters is that the implementors know the
"philosophy" of the standard, and that they should try to follow it as much as
possible. Anyway, we won't obtain hardest to round cases for all
transcendentals: there will remain functions for which we won't know if we can
round them correctly  in a reasonable time.


Best regards

Jean-Michel


754 | revision | FAQ | references | list archive