Re: [Bug-apl] IOTA

[Juergen Sauermann]

Hi,

I have changed GNU APL so that B ≡ ⊂B if and only if B is a simple scalar.
Before the change B ≡ ⊂B also for non-simple scalar B.

Unfortunately Elias' correct observation that (⊂⊂B) ≡ ⊂B in GNU APL was almost
an axiom in GNU APL that nows has turned out to be wrong. That - now incorrect -
axiom was enforced (or assumed to be valid) in several places in GNU APL and it
could be that other functions (or, more likely, operators) are affected as well.

Please let me know if you discover other cases where (scalar and non-simple) values
are being incorrectly enclosed or incorrectly not enclosed. SVN 699.

/// Jürgen

On 03/03/2016 01:26 PM, Juergen Sauermann wrote:

Hi,

I see. I guess I overlooked the 'simple' attribute in both definitions.

I will fix this in GNU APL.

/// Jürgen

On 03/03/2016 12:53 PM, Kacper Gutowski wrote:

On Thu, Mar 3, 2016 at 12:40 PM, Juergen Sauermann
<address@hidden> wrote:

the IBM APL2 language reference says (page 111):

If R is a simple scalar, ⊂ R is R. If R is not a simple scalar, the depth of
R is 1+ ≡R.

And the ISO standard says the same (page 169):

Z ← ⊂B
 Note: If B is a simple-scalar, Z is B.

Jay is nevertheless correct, because ⊂x is NOT a simple-scalar
according to ISO.  On page 21 it has the following definitions:

Simple: An array is simple if each item of its ravel-list is either
  a character or a number.
Simple-scalar: A simple array whose rank is zero.

The sole item of ⊂x is an enclosed array, not a character or number.


-k