Re: [Json] REMINDER - WGLC Ends 2013-10-11

[John R Levine <[e-mail]>]

> I’m OK with John’s changes, except for the MAY variation in section 9.  The
> second half, with all the instances of lower-case “may”, is talking about
> implementation behavior.  I had read it as advice for the reader as for
> what implementations might do.  So turning those into MAY means that they
> normatively describe behavior of conforming implementations.  Are we really
> OK with that?  My instinct is that, since this spec has held together
> pretty well all these years, why don’t we leave this alone?  -T

There is a somewhat theological argument that if you use 2119 language at 
all, the capital and lower case versions of the magic words are equally 
normative.  You could avoid that by writing around it, e.g.

    An implementation might set limits on the size of texts that it
    accepts.  An implementation might set limits on the maximum depth of
    nesting.  An implementation might set limits on the range and precision
    of numbers.  An implementation might set limits on the length and
    character contents of strings.

R's,
John

PS: There's another argument that since informational RFCs aren't 
normative, 4627 shouldn't have used 2119 language at all, but that horse 
is long since water under the barn.