[Jprogramming] Bug: +./4.57 4.34 4.44

[Hauke Rehr]

The more I think about it, the more I think +. could actually
be amoung those arithmetic primitives that would benefit from
using 0&= without doing much harm if I don’t miss something.


Still otoh:
The decimal system is an arbitrary way of representing numbers.
There’s no point in having 0.142 be tolerantly 0.14 in base 10
rather than tolerantly 0.1 in base 7. How would you know?
That’s what rationals are made for. No loss of precision.
Whatever the base. And then there are symbols and symbolic
expressions – but that’s a different story altogether.


Am 15.05.20 um 23:37 schrieb Raul Miller:
> On Fri, May 15, 2020 at 11:20 AM 'Bo Jacoby' via Programming
> <programming at jsoftware.com> wrote:
>> =  is tolerant, why isn't   +.    ?
> 
> For the same reason that - is intolerant.
> 
> Perhaps we should have a tolerant (-) but how would that work?
> 
> Would it be fundamentally different from multiplying by an integer
> constant, using floor to guarantee integer representation, performing
> the sensitive operation and then dividing by that same constant?
> 
>     e_2=: &:<.(&.:(*&100))
>     +./e_2   4.34 4.44 4.57
> 0.01
> 
> Thanks,
> 

-- 
----------------------
mail written using NEO
neo-layout.org