RE: [xri] XRI Names and XRI Addresses

["Drummond Reed" <drummond.reed@onename.com>]

Good points, Gabe. Three responses:

1) It's a good question which feature of XRI-URNs, aka XRI addresses, is
most notable. I'd been assuming it is their "URN-ness", i.e., their
persistence. But with XRIs, it may be that their efficient machine
routability is more important. Without question both are key features.

2) You refered to the word "permanent" but I used the word "persistent"
(which is also the term used in RFC 2141). I'm suggesting we should even
change our glossary entry from "permanent identifier" to "persistent
identifier" simply because "permanent" implies that the identifier will
always work, which might be nice, but is unfortunately not something XRI
addresses (or any other URN system) can enforce. The responsible URN
authority can always stop functioning, or, in another scenario, a
resource controller may for privacy reasons chose to "abandon" a
particular URN (for example, "URN x") and start using another one (say
"URN y"). Although URN x will never be reused, it seems counterintuitive
to use the word "permanent" to describe this scenario.

(In truth, the best overall term might be "non-reassignable" because
that's all a URN really is, but the term "persistent" is already very
widely associated with URNs.)

3) I especially appreciate your point that the term "persistent address"
highlights the fact that the term "XRI address" could also refer to a
non-persistent address, i.e., an XRI name. It's ironic then that this
problem was the original source of my suggestion. The term "address" is
so widely used to refer to identifiers of any kind (persistent,
reassignable, semantic, non-semantic, absolute, relative, etc.) that I'm
worried that the term "XRI address" will be taken by most readers to
mean "any XRI" rather than just XRI-URNs. The analog, for example, is
that the term "Web address" is widely used to refer to both URIs based
on DNS names and any other form of URI, including URNs.

I don't have a good solution to this problem, however - I stewed on it
all weekend. I really like the term "XRI name" for XRI-URIs because it
is a good analog to DNS names. But I wish we could find a more precise
human-friendly term for an XRI-URN than "XRI address".

Ideas?

=Drummond 


-----Original Message-----
From: Wachob, Gabe [mailto:gwachob@visa.com]
Sent: Monday, May 05, 2003 3:46 PM
To: 'xri@lists.oasis-open.org'
Subject: RE: [xri] XRI Names and XRI Addresses

Does this beg the question of what XRI Non-permanent addresses are? By
adding the modifier in there, you imply that this phrase is a subset of
a larger set (ie addresses, both permanent and non-permanent).

I'm not totally convinced the use of permanent in the middle is more
confusing than helpful.

If you want to connote permanentness, ID is the correct phrase (as had
been in XNS). If you want to connote machine routability, then address
is probably the correct phrase.

You want to suggest that XRI URNs are both. I'd argue that the routing
part is slightly more important if you had to focus on emphasizing one
feature over the other (ie without the modifier in the middle). This is
because the routability/resolvability is a unique feature in the URN
space, whereas permanence isn't.

So I guess I'm -1 on the proposal to add "permanent" in the middle.

        -Gabe

> -----Original Message-----
> From: Drummond Reed [mailto:drummond.reed@onename.com]
> Sent: Monday, May 05, 2003 11:37 AM
> To: xri@lists.oasis-open.org
> Subject: [xri] XRI Names and XRI Addresses
>
>
> I had an action item to raise the issue in the v4 Requirements doc of
> the introduction of the terms "XRI Name" and "XRI Address". In this
> draft we only used these terms in the first half of the Motivations
> section, but in the final draft (and most importantly, in the spec
> itself), we should use them consistently throughout.
> 
> These terms were proposed by Dave Wentker as "human-friendly
> identifiers" for what we have been referring to technically
> as XRI-URIs
> and XRI-URNs. The reason is that with coming of a formal spec, we will
> need to begin communicating with a much larger audience about
> the major
> features of XRIs, and one of them is that XRIs provide a
> standard syntax
> for both reassignable, human-friendly abstract identifiers and
> persistent, machine-friendly abstract identifiers. So we need clear,
> consistent, understandable terms for both layers.
> 
> Please post any thoughts you have about standardizing on these terms.
> 
> My one suggestion after chewing on this over the weekend is that we
> might want to use the modifier "persistent" when talking about XRI
> addresses because that is what so significantly distinguishes
> them from
> IP addresses (or any form of name, whether a DNS name or an XRI name).
> So from an Internet resolution standpoint the four possible
> layers would
> be:
> 
> * XRI name
> * XRI persistent address
> * DNS name
> * IP address
> 
> You can skip a layer any time a resource doesn't need the features of
> that layer, i.e., an XRI name can refer directly to a URI or an IP
> address if the resource does not need a persistent
> identifier, or an XRI
> persistent address can refer directly to an IP address if it doesn't
> need a DNS name.
> 
> =Drummond
> 
>