Re: [Bug-apl] IOTA

[Juergen Sauermann]

Hi Mike,

I see. What is important  for GNU APL is compatibility with IBM APL2,
whatever the reason might have been.

I could also imagine that enclosing a simple scalar would give a nested simple scalar,
that would also satisfy the condition of disclose and enclose being inverses of each other,
but the implementation choice made in APL2 was apparently a different one.

/// Jürgen

On 03/06/2016 05:08 PM, Mike Duvos wrote:

Hi Jürgen,

When James A. Brown wrote APL2, he based his arrays on Trenchard
More's "Array Theory", an attempt to give nested rectangular arrays an
axiomatic foundation equivalent to that of set theory.  The major
implementation of this idea is in the language NIAL, of which Q'NIAL
is a popular incarnation.

I haven't read More's paper, but I would assume that for array theory
to be consistent, enclose of a non-simple scalar can't be a no-op,
because disclose of a non-simple scalar isn't a no-op, and enclose and
disclose need to be inverses of each other.

Regards,

Mike