Minutes for URNBIS at IETF-90
minutes-90-urnbis-1

URN BIS Working Group Minutes
2014-07-25 1150-1320
Chair: Andrew Newton
Minute taker: Dorothy Stanley, Aruba Networks
Agenda:
IETF 90 URNBIS Agenda
Chair: Andrew Newton

1. Agenda Bashing
2. WG Status
3. How do query and fragment components relate to URNs? Sean Leonard (Penango,
Inc) 4. draft-ietf-urnbis-urns-are-not-uris -  John Klensin 5. Any Other
Business

The chair opened the meeting, bringing to the attention of all participants the
NOTE WELL. He then showed the agenda, noting a slight change from the published
version, and asked if there were any requested changes. None were offered. The
chair then reviewed the status of the working group. He noted that the working
group was stalled, and the current debate centered around moving URNs out of
the URI definition.

Sean Leonard presented his draft, draft-seantek-certspec (see slides). There
was then a discussion around query components. Keith Moore stated that he
thought queries would be useful in URNs, and that he believed their use was in
scope for the original intent of URNs. Joe Hildebrand noted that his use of
URNs in software and protocols is where he does not expect them to be parsed.
The chair then asked Joe if URIs could be used in places where he uses URNs,
and he said yes but the use of URNs in these situations is so they won’t be
parsed. The chair then asked how equivalence is accomplished if the software
doesn’t parse URNs. Sean discussed his thoughts on fragments, and the group
discussed fragments as they relate to semantics. Leslie Daigle noted that the
original URN working group made no progress when discussing semantics and only
achieved their goals after focusing only on the specification. Larry Masinter
then asked the chair about the scope of the work, and specifically asked if
discussing the original intent of the original URN working group was part of
it. The chair gave his personal opinion that the output of the working group
should be break as few things as possible regardless of original intent. The
group then discussed how URNs are used, with Joe noting that a lot of software
using URNs do not parse them. The chair asked if XML namespace-aware parsers
are a good example, and Joe agreed. The group discussed how the comparisons
worked in those cases. There was then a discussion of queries and fragments
being scheme dependent. Keith Moore stated that he believed it would be a bad
idea to make queries and fragments scheme dependent. John Klensin presented his
draft, draft-ietf-urnbis-urns-are-not-uris, next. However, he did not present
the slides that had been previously distributed. He noted that the previous
discussion and discussions he had had on the mailing list and with individuals
recently illustrate how his draft came about . He noted that the group could
have endless discussions about the nature of URNs, conformance with RFC 3986
and a myriad of other things. He speculated that there are two categories of
URI parsers, ones that look at the scheme separated by the colon and process
specifically on the scheme, and ones that try to process components of a URI.
He noted that nothing the group can do with URNs will impact the first type,
but that the second type might be impacted in a limited way. Joe Hildebrand
proposed a thought experiment to move the discussion forward where the group
could discuss a new URI scheme, how to use strings within that scheme, and to
avoid using the # mark. Leslie Daigle stated that she didn’t “hate” that idea.

Larry thought the group would quickly devolve into the philosophy of names vs
locators and other issues such as permanence vs resolution, disputes, etc..
Leslie stated that would leave the possibility for differing opinions on URN
definitions. Keith agreed with Leslie and Joe.

The chair then focused the discussion on the differing opinions of fragments.
Larry noted that fragment identifiers are for URLs and the specifications don’t
match how they are used, and one such issue with them are for equivalence. John
stated that the group had spent 2 years discussing fragments and their meaning
and should instead focus on the necessity of strings starting with a # sign and
the syntax for them. Keith argued that fragments are a mess and should not be
extended to URNs. Sean also stated he believed fragments should stay out of
URNs.

Dave Thaler then turned the discussion back to Joe’s proposal. He noted that it
would cause issues with APIs and name spaces. Joe stated that the intent of his
proposal is to provide “legal URIs” that do not need to be heavily parsed. Ted
Hardie offered a compromise suggesting that the group not start with the
assumption that both are URIs stating that some needs might be met with legal
URIs and others not. Joe stated that another requirement he envisioned was a
reasonable amount of uniqueness.

Larry then began a discussion of the needs of the library community, noting
that they were not participating in this discussion. John stated that he
believes that community thinks it is knowledgeable about URNs and need queries
and fragments, noting that three years ago they wanted to standardize queries
and fragments in URNs outside of the IETF. Larry pointed out that a subset of
that community went to OASIS to standardize XRIs for their needs. The chair
then asked if this was a similar to approach to Joe’s, upon which Joe said he
believed it to be so. The chair then asked if the library community would be
opposed to a new scheme. John stated he believed they want a parsable
identifier, not a pure identifier. Joe stated that his solution is about pure
identifiers rather than parsable identifiers.

Julian Reschke then asked the group to bring the discussions back to the
question of URNs being part of URIs and being a URI scheme. He asked for
examples of identifiers that can’t fit in a URI. Ted noted if a design
constraint is to conform to URIs, that might be taking some options off the
table. John stated that URN conformance to URI syntax was not a problem but
that the disagreements are over semantics. Julian disagreed. Keith stated that
he was not happy with RFC 3986, and noted his desire to have a document that
did little more than URI scheme registration.

The group then discussed compatibility issues of URNs and URIs. The chair noted
his own experience with an IPv6 study that found various URI libraries did not
handle IPv6 addresses well. Julian stated that if URNs are separated from URIs,
then URNs could no longer be used in all the places that specify URIs
generally. John suggested the group start with the same syntax but use
different semantics. Larry called for a BCP on how to read RFC 3986 and RFC
2141. John stated that was not part of the groups charter.

The group then discussed moving forward with looking at URN syntax while trying
to keep compatibility with URI syntax. The chair and Barry Leiba, the group’s
Area Advisor, then discussed using John’s draft as a tool for guiding the
contents  of the 2141bis document, stating that  John’s document may be
published as an RFC or it may not. There were no objections from the group on
moving forward in this manner.

The chair then thanked all participants and closed the meeting at 13:25.