OASIS Mailing List ArchivesView the OASIS mailing list archive below
or browse/search using MarkMail.

 


Help: OASIS Mailing Lists Help | MarkMail Help

xri message

[Date Prev] | [Thread Prev] | [Thread Next] | [Date Next] -- [Date Index] | [Thread Index] | [List Home]


Subject: RE: [xri] Comments on requirements draft-02 motivations


Gabe, great feedback. I'll incorporate it into the next draft that I'm doing
on Wednesday.

I think the distinction between level 5 - URLs - and level 6 - URNs is the
most fundamental point in the whole piece, so I'll try to make sure it's as
clear as possible. I agree that the label "URL" is probably something we
should avoid, since it has become so nebulous. I'm tempted then to label the
levels using our glossary terms as follows:

Layer 7 - Reassignable Abstract Resource Identifiers
Layer 6 - Permanent Abstract Resource Identifiers
Layer 5 - Concrete Resource Identifiers
Layer 4 - DNS Names
Layer 3 - IP Addresses
Layer 2 - MAC Addresses
Layer 1 - Internal Machine Addresses

Layers 5, 6, and 7 are all URIs, just for different purposes and with
different requirements. The whole reason for XRIs is to have an
interoperable standard for abstract URIs - layers 6 and 7.

You asked about Layer 1 - it is the internal address of a resource within a
particular MAC-addressable machine, i.e., a filename, a spreadsheet cell, a
database record, etc. Layer 1 addresses are allocated by operating systems
and applications. Does that answer your question?

=Drummond 

-----Original Message-----
From: gwachob@visa.com [mailto:gwachob@visa.com]
Sent: Sunday, March 30, 2003 8:28 PM
To: xri@lists.oasis-open.org
Subject: [xri] Comments on requirements draft-02 motivations

Drummond-

Comments on the requirements motivations section. Overall its fine - here'
some specific comments.

  -Gabe


line 83: whats the "layer 1"? I'm not sure what you mean here. Whats the
difference between linking and naming? Also, the stacking of URN on top of
URL is perhaps contentious. The offical "modern view" is that URNs and URLs
are mostly non-overlapping and they are both subsets of the space of URIs.
URLs are actually not formally defined anywhere (some people think HTTP URIs
when you say URLs, some think its exactly the same as URIs, some think its
some category of URIs that have a machine address in them)

line 88-89: the description of IP should talk about "between other
machine-level networking protocols" - probably better to say "between local
networks"

line 99-102: URNs are NOT web-resolvable URLs. Some argue that the
non-resolvability of URNs is whats held them up from being used. Us
providing at least a basic resolutoin mechanism will make

line 116-122: I think you aren't addressing the real problem with URNs (or
I'm not following you). I'm not sure the problem is best stated as that they
are hard to use for humans. I think the problem you are getting at is that
there is an inherent tension between the persistence of URNs and the
transitivity of human understanding of terms. There are certainly
non-transient things that can be identified by URNs (ie a particular SENSE
of a word - like "bare" can mean "undressed" or it can mean "plain"). So the
problem is that the symbols humans are used to have inherent non-persistence
and "imprecision" that are in conflict with the precision and permanence of
URNs.

lines 143-148: the "postal mail to the bathroom" example is a little
strained. Perhaps the concept of addressing to a corporate role (ie
"President") is more illustrative (ie to addess mail to "President, XYZ
Corp, 123 Main St, Anywhere, CA, 94001)


[Date Prev] | [Thread Prev] | [Thread Next] | [Date Next] -- [Date Index] | [Thread Index] | [List Home]