Re: [Json] A minimal examplotron-style JSON validation language.

[John Cowan <[e-mail]>]

On Wed, May 29, 2019 at 5:15 PM Carsten Bormann <[e-mail]> wrote:


> While it is certainly enjoyable to design another language,


I cheerfully admit to that motive.


>  I’m not sure why this should not be done in an ABNF-like syntax

such as CDDL (which, by the way, is in AUTH48 and will soon be RFC 8610).

Keeping your names (even the snake case, where I’d be using kebab case):
>

My effort with this schema language is not to show that it can do something
that other proposals cannot, but that it is the simplest thing that can
possibly
work for both dynamically and statically typed programming languages,
and that everything else can and should be left to higher (inner) layers
that do general-purpose validation.  (Another example: a field that
must be a "dictionary word" in the sense that it appears in a particular
dictionary, whose contents may change over time.  Implementing this
in the schema language will require a database lookup capability oe
perhaps the ability to call a web-services API.)

 Antoine St.-Exupery said: "A designer knows he has achieved perfection
not when there is nothing left to add, but when there is nothing left to
take away."
("Il semble que la perfection soit atteinte non quand il n'y a plus rien à
ajouter,
mais quand il n'y a plus rien à retrancher.")  That's what I'm striving for.
My design has the advantage of being easy to implement in any programming
language, even Fortran.  It is also "obvious", with nothing to debate
except the syntax,
because the answer to "Shall we include this or that feature?" is almost
always "No."

Another amendment:  [] is the type definition of a polymorphic array, as {}
is the
type signature of a polymorphic object.