preANN: ArbFloats. the sequel. (extended precision *values* as floating point numbers)

[Jeffrey Sarnoff]

If the first ArbFloats.jl was proof of concept, this one is concept of proof.

This is a pre-announcement in response to some requests for access and my preference.

These quantities are information-richest floating point representations of using arithmetic, elementary and some special functions.

ArbFloats presents digits that neither lead nor mislead the perception of value, showing more or fewer digits would be less informative.

When announced (ANN), I will offer more explanation and examples and any overlooked details, 

no worries -- if you do not need this now, wait for me to have fuller tests written, etc.

.

If you do need it or just want to play with it, please do and share your experience with me.

https://github.com/jsarnoff-juliacon/ArbFloats.jl

[Tom Breloff]

Thanks Jeffrey.  Starred and followed.  For those following from a distance, ArbFloats (IMO) are a much faster alternative to "fixed-precision Unums 1.0", and could nicely form the basis of interval arithmetic with open/closed endpoints, paving/uboxing algorithms, and more.