Re: URN Namespace for MEF

[Mahesh Jethanandani <[e-mail]>]

Dale,

First of all, thanks for a detailed review of the draft. 

Comments/questions inline.

> On Nov 6, 2015, at 2:57 PM, Dale R. Worley <[e-mail]> wrote:
> 
> Section 2:
> 
>   Declaration of syntactic structure:
> 
>      The Namespace Specific String (NSS) of all URNs that use the MEF
>      NID will be maintained by MEF Assigned Names and Numbers (MANN)
>      registry.
> 
> This doesn't specify the syntactic structure.  I assume that the
> intention is that this registration is to allow MANN to assign any NSS
> that would be allowed by the generic URN specification.

That is correct.

>  But you should
> say that.  I think you could say
> 
>      The syntax of namespace specific strings for the 'mef' namespace
>      is <NSS> in RFC 2141.

Ok.

> 
> The statement you now have in the section describes who manages mef
> URNs, which is a useful thing, but doesn't belong in this section, I
> think.  But I can't spot the place in the registration template where it
> should go.

> 
>   Identifier uniqueness considerations:
> 
>      MEF could allow for use of experimental type values for testing
>      purposes only.  Note that using experimental types may create
>      collision as multiple users may use the same values for resources
>      and specific strings.
> 
> First, you should say "for *different* resources ...".  If multiple
> users use the same value for the same resources, there is nothing
> unusual about that.
> 
> This provision doesn't bother me, but is likely to be objected to by
> others.  And there can be practical problems -- experimental uses of
> protocols by one user sometimes "leak" into experimental environments
> operated by another use, causing unexpected results.  You should give
> some though to avoiding this, IMO.  Perhaps:
> 
> 1) Devise a system for allocating experimental values that allows each
> user to generate values that will not be allocated by other users.
> E.g., the pattern "urn:mef:experimental:foobar.com:..."; or

That is the idea. The current thought is that all experimental values will have a pattern of “urn:mef:xid:…” where xid refers to experimental id.

> 
> 2) Define the "value" of an experimental URN in terms of its *function*
> or high-level *meaning* ("denotes the string to be returned when
> conditions X and Y occur"), allowing different users to have it be
> translated in different manners or cause different effects, but always
> with the same higher-level significance.

Thanks.

Mahesh Jethanandani
[e-mail]