Re: [urn] Registration request for urn:eris: namespace

[Peter Saint-Andre <[e-mail]>]

On 10/3/23 9:44 AM, Dale R. Worley wrote:
>> To restate what I meant in a way that hopefully is clear, what I don't
>> like about the current proposal is that the identifiers aren't
>> *location-independent*.
> 
> To flesh out some aspects of this:  On one hand, I'm objecting that the
> design of the namespace doesn't allow, doesn't conceptualize, that the
> URNs are location-independent names.  On the other hand, I'm proposing
> ways that the namespace can reserve a subspace for identifiers that,
> indeed, aren't location-independent names.
> 
> This is like what happens with IP addresses.  IP addresses, of course,
> are globally unique.  Except for 10.0.0.0/8 addresses et al., which only
> have a meaning within a local context that defines it.
> 
> The current Eris proposal, both in concept and in details, is like the
> identifiers passed around in all distributed file systems:  "filesystem
> ID"/"file ID"/"block offset".  None of those consider their identifiers
> to be *names*.  But while I want y'all to reconceptualize the namespace
> as names and add details to support that, I don't want you to have to
> adjust your experimental implementations to have to support real names
> yet.

Dale, I think what you and I are saying is threefold:

1. URNs aren't a great fit for the Eris system as currently designed.

2. Although the Eris designers *might* want to modify their system to 
conform to URN requirements, that's their decision and we're by no means 
insisting on it.

3. However, absent such a redesign, it seems better to use something 
other than URNs in Eris.

Peter