Re: [Bug-apl] Integer - rational reciprocal anomaly

[Juergen Sauermann]

Hi Fred,

thanks, fixed in *SVN 998*.

I also changed the alignment of quotients in columns
from align to E to align to decimal dot. The previous
example from Ala'a now looks like this:

*      ⎕PS ← 1 0**
**      n,⍪{+/1÷3⋆⍳⍵}¨n←⍳40**
** 1                   1÷3**
** 2                   4÷9**
** 3                  13÷27**
** 4                  40÷81**
** 5                 121÷243**
**                     ...**
**37  225141952945498681÷450283905890997363**
**38  675425858836496044÷1350851717672992089**
**39 2026277576509488133÷4052555153018976267**
**40                   0.5*

/// Jürgen


On 08/16/2017 05:17 PM, Frederick Pitts wrote:
> Hello all,
>
>         Using SVN 996, configured with RATIONAL_NUMBERS_WANTED=YES, I
> observe the following:
>
>        ⎕PS ← 1 22
>        ÷ ¯2
> ╔══════════════════════╗
> ║1÷18446744073709551614║
> ╚══════════════════════╝
>        1 ÷ ¯2
> ╔════╗
> ║¯1÷2║
> ╚════╝
>        ÷ 2
> ╔═══╗
> ║1÷2║
> ╚═══╝
>        1 ÷ 2
> ╔═══╗
> ║1÷2║
> ╚═══╝
>         The first result above is wrong.  Additionally, taking the
> reciprocal of any other negative integer yields an incorrect result.
>
> Regards,
>
> Fred