Feature request: toBiggestInt

[ehmry]

I need to convert BigInt to float64 and I'm not sure I'm qualified to implement it.

I'm using this:

proc toBiggestFloat(a: BigInt): BiggestFloat =
  for i in countdown(a.limbs.high, 0):
    result = result * BiggestFloat(1 shl 32) + a.limbs[i].BiggestFloat
  if Negative in a.flags:
    result = - result

But it gives me different results from a Haskell implementation as it get close to saturating the float at 179769313486231560835325876058105298516207002341652166261661174625869553267292326574530099287946549246750631490335877017522087105926987962906277604735569213290190919152394180476217125334960946356387261286640198029037799514183602981511756283727771403830521483963923935633133642802139091669457927874464075218945.

[pietroppeter]

What is the Haskell implementation you are refering to? I suspect that it might support arbitrary precision floats. In fact the value that you report for saturation it seems very close to highest possible value for a float64 (e.g. from here):

1.7976931348623157e+308

Regarding your implementation it seems ok, although it might not be the most efficient since for big numbers the latest addition might not change the float much. Looking around I found a nice discussion in Julia (summary: you just need the highest 54 bits of precision since this is what a float64 can represent) where a more performant implementation is proposed and later merged. We could take inspiration for that in order to put a toBiggestFloat in bigints.

update: maxFloat64 in nim

from the hex representation of a max float64 wikipedia, here is a way to echo max float64 in nim (playground):

import strutils
echo cast[float64]("0x7FEFFFFFFFFFFFFF".parseHexInt)

output:

1.797693134862316e+308

Note that float64.high gives inf.