On 3 November 2017 at 17:09, Poul-Henning Kamp <phk@phk.freebsd.dk> wrote:
> --------
> In message <CACweHNBjCu-Ymm7OX70u65QR-ws_HKY6U7qcCJR8Y4AWzuPD1Q@mail.
> gmail.com>
> , Matthew Kerwin writes:
>
> >If 'Foo' is an integer known/defined to be within [1,3600], say, then its
> >spec can say: "the value MUST be a <theoretical-unsigned-4digit-type>
> >between 1 and 3600, respectively".
>
> The entire reason behind this draft is that the spec-writer can
> write:
>
> "The value MUST be 1 <= value <= 3600, and serialized as a
> Common Structure Number [RFCXXXX]."
>
> and *not* write any ABNF at all.
>
>
It might be my poor use of language, but I was trying to write almost
exactly what you did: "The value MUST be 1<=value<=3600, and serialised as
a Common Structure
Whatever-We-Call-The-Short-Number-If-There-End-Up-Being-Multiple-Number-Types
[RFCXXXX]."
My angle brackets weren't meant to signal ABNF elements.
Cheers
--
Matthew Kerwin
http://matthew.kerwin.net.au/