Re: Gsoc Rejection, or SUMMER OF COD

[Ori Bernstein <ori@xxxxxxxxxxxxxx>]

On Tue, 13 Feb 2018 20:58:31 -0500, "S. Gilles" <sgilles@xxxxxxxxxxxx> wrote:

> I'd like to formally apply for working on libm. Muller et al.'s
> Handbook of Floating-point Arithmetic appears to be incredibly
> detailed, and the numerous libraries from other projects should
> provide good test data.

You may find yourself looking at Sun's fdlibm over and over :)

> It probably won't be the flashiest or
> slickest collection of algorithms around, but hopefully it will be
> respectable.
> 
> -- 
> S. Gilles

Respectability is perfect -- easy to understand and follow beats
slick algorithms every time, in my books.

Since libm is largely a bunch of standalone functions, I'd tackle it
one function (family?) at a time -- when you get an implementation
you're happy with, send in a patch to lib/math, and over time we end
up growing it to cover whatever scope we feel is appropriate.

I figure you've got the judgement to know in which order to tackle
them.

-- 
    Ori Bernstein