Re: [Ltru] Re: Informative (was: Extended language tags)

Hi -

> From: "Doug Ewell" <dewell@roadrunner.com>
> To: "LTRU Working Group" <ltru@ietf.org>
> Sent: Tuesday, October 09, 2007 6:56 AM
> Subject: [Ltru] Re: Informative (was: Extended language tags)
...
> I must not be a good enough RFC 2119 language lawyer to understand this. 
> I thought MUST was normative, and SHOULD and MAY were non-normative.

No.  If language is relevant to determining whether an implementation
conforms to a specification, it is normative.  Specifying something as
an option does NOT make it non-normative.  In general, "informative"
material is stuff that can be removed from a specification without
changing any implementation requirements.  "SHOULDs" and "MAYs"
obviously do not fall into that category - they are normative language.

> It's hard for me to see how "the full implications must be understood 
> and carefully weighed," in the RFC 2119 definition of SHOULD, can be 
> applied in a normative sense.

The fact that a particular feature is OPTIONAL doesn't make that part
of the specification "informative".  Think snow tires.  They may be
optional in some areas, but when you do get them, you want them to
fit your car's rims and behave as specified.

> I note that RFC 4646 contains a lot of MUSTard, but almost no direct 
> references to fields or behaviors being "normative" or "informative."

"informative" material should be used sparingly in a specification,
because folks have an unfortunate tendency to confuse an illustration
or example with the specification itself.

Randy

_______________________________________________
Ltru mailing list
Ltru@ietf.org
https://www1.ietf.org/mailman/listinfo/ltru