Re: [apps-discuss] "X-" revisited

[Paul Hoffman <[e-mail]>]

On Jul 6, 2011, at 2:01 PM, Peter Saint-Andre wrote:

> On 7/6/11 2:53 PM, Dave CROCKER wrote:
>>  Authors of application protocol specifications SHOULD provide
>> extensible registries for all parameters and SHOULD mandate use of the
>> registries, for all values of the parameters, independent of the form of
>> the parameter names.
> 
> The second SHOULD strikes me as somewhat controversial. :)

Not much more controversial than the whole topic. I can quickly name many protocols for which "some values are reserved for experimental use" have had horrible problems in practice, either where an experimental value ended up being baked into shipping products or two vendors used the same experimental use value for different things.

Maybe the Apps Area is not as familiar with "values 250 through 255 are reserved for experimental use" than the rest of the IETF, but the topic is quite germane to this draft. Dave's proposed wording is hard to swallow, but probably best for the real-world long-lived Internet.

> The first SHOULD is fine by me, although I'm not sure what an extensible
> registry is -- did you mean "both permanent and provisional registries"?


And extensible registry is one where there is a registration procedure so that today's registry might be larger tomorrow.

--Paul Hoffman