RFR: 8077587: BigInteger Roots [v19]

[Hans Boehm]

Apologies if this has been discussed before.

The explanation for the exception for negative n ("This would cause
the operation to yield a non-integer value.") feels wrong to me. I think
the definition of the result as: x.signum() * floor(abs(nthRoot(x, n)))
works fine for that case, if nthRoot(x,-n) is 1/(nthRoot(x,n)) as expected,
right? And the main definition of the result just isn't well-defined.
Neither yields a non-integer.

AFAICT, the two options here are:

1) Just produce the, usually zero, result, usually with a remainder of
`this`. (If 'this' is zero, you would still need an arithmetic exception,
and +-1 is also interesting.)

2) Produce an ArithmeticException with a justification along the lines of
"The result would not be meaningful."

Based on the Wikipedia page, negative n aren't normally a thing, so (2)
seems justifiable. But I think the current justification for the exception
is not.

I can't think of a good reason to prefer (1), but we've all been bitten by
unnecessary domain restrictions, so I'm not positive about that.

Hans

On Fri, Jul 11, 2025 at 7:42 AM fabioromano1 <duke at openjdk.org> wrote:

> > This PR implements nth root computation for BigIntegers using Newton
> method.
>
> fabioromano1 has updated the pull request incrementally with one
> additional commit since the last revision:
>
>   Optimize the computation of the input's shift
>
>   Optimize the computation of the input's shift, in order to avoid having
> more non-significant bits than necessary in the initial estimate.
>
> -------------
>
> Changes:
>   - all: https://git.openjdk.org/jdk/pull/24898/files
>   - new: https://git.openjdk.org/jdk/pull/24898/files/788a82b9..6dcd6792
>
> Webrevs:
>  - full: https://webrevs.openjdk.org/?repo=jdk&pr=24898&range=18
>  - incr: https://webrevs.openjdk.org/?repo=jdk&pr=24898&range=17-18
>
>   Stats: 18 lines in 1 file changed: 8 ins; 0 del; 10 mod
>   Patch: https://git.openjdk.org/jdk/pull/24898.diff
>   Fetch: git fetch https://git.openjdk.org/jdk.git
> pull/24898/head:pull/24898
>
> PR: https://git.openjdk.org/jdk/pull/24898
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://mail.openjdk.org/pipermail/core-libs-dev/attachments/20250711/2d332945/attachment.htm>