Re: [Bug-apl] Outer product with replicate error

[Juergen Sauermann]

Hi Peter,

is seems like IBM APL2 computes

      1 2 3 /¨ 4 5 6

as:
      (1 2 3/4) (1 2 3/5) (1 2 3/6)

However, the ISO standard says

Z ← A f¨ B
Informal Description: The operand function f is applied independently to corresponding
items of the arguments, or (monadic case) independently to the items of the argument.
The corresponding results are assembled in an array of the same shape as the argument(s).

Evaluation Sequence:
...
For form Af¯B,
Set X to item I of the ravel-list of A.
Set Y to item I of the ravel-list of B.
For each pair X and Y of corresponding items from A and B, Evaluate-Dyadic-
Function with XfY giving token T.
...

I believe this is one of the very few cases where IBM is wrong.

/// Jürgen

On 05/16/2016 04:25 PM, Hans-Peter Sorge wrote:

Hi,

just to mention, there is an other difference in interpretation
regarding the scan operator GNU-APL vs IBM-APL.

GNU-APL:
      1 2 3 /¨ 4 5 6
 4  5 5  6 6 6

IBM-APL:

           1 2 3 /¨ 4 5 6
 4 4 4 4 4 4  5 5 5 5 5 5  6 6 6 6 6 6


Best Regards,
Hans-Peter