Christopher Kormanyos
Christopher Kormanyos
New insight has been gained in the area of Karatsuba square root. This is currently under investigation in another project and should be applicable to this issue eventually.
Karatsuby square root is being investigated more deeply in Boost.Multiprecision
> Assuming that it is a bug to pass a negative value to a `sqrt` function, might it be best to assert this as a precondition In wide_integer, this has,...
> bugs are treated with zero tolerance Fair enough. It's always a toss-up, provide the wrong answer or throw. But wide-integer is designed to be on-the-metal, no-rtti, exceptions off. I...
> HTH and thanks for asking the question. It really did. Thank you @johnmcfarlane > I have been planning to write an article and maybe even present a talk on...
> call those out and invite you to review I'd like that. Thx.
For post-Boost1.80 simply eliminate exceptions (and RTTI) with the compiler options `-fno-exceptions` and `-fno-rtti`.
> For post-Boost1.80 simply eliminate exceptions (and RTTI) See the working model for this in the [wide-decimal project](https://github.com/ckormanyos/wide-decimal).
A lot of the missing lines in those approximately 3% of missing code coverage are used in I/O streaming, in particular with octal and hexadecimal formats and some less common...
See also #2 and this link to [128-bit division](https://danlark.org/2020/06/14/128-bit-division)