Re: Formatting numbers without precision loss

[Lorenzo Donati <lorenzodonatibz@...>]

Very interesting!

I duckducked for it (I'm trying to dump google :-) and found the 
relevant article (luckily not paywalled):

https://dl.acm.org/doi/pdf/10.1145/3192366.3192369

I didn't know about that. As I said in another post, it's been a very 
long time since I looked into these problems.

Thanks for the tip!

-- Lorenzo

P.S.: I also found a git repo from the article's author (Ulf Adams):

https://github.com/ulfjack/ryu

On 21/06/2023 00:02, Andrew Gierth wrote:
> > > > > > "Lorenzo" == Lorenzo Donati <lorenzodonatibz@tiscali.it> writes:
>
>   Lorenzo> and also this SO post:
>   Lorenzo> https://stackoverflow.com/a/2174100/2633423
>
>   Lorenzo> which links to the article above and other info.
>
> That's well out of date; the Ryu algorithm, published in about 2018,
> beats the pants off every other exact float output algorithm both in
> speed and simplicity.
>
> (It's hard to overstate how good it is. It's many times faster than any
> previous algorithm, requires no special cases other than zero/inf/NaN,
> and doesn't use any float arithmetic at all.)