Internet DRAFT - draft-ietf-urnbis-urns-are-not-uris
draft-ietf-urnbis-urns-are-not-uris
Uniform Resource Names (urnbis) J.C. Klensin
Internet-Draft July 4, 2014
Updates: 3986 (if approved)
Intended status: Standards Track
Expires: January 03, 2015
Names are Not Locators and URNs are Not URIs
draft-ietf-urnbis-urns-are-not-uris-01.txt
Abstract
Experience has shown that identifiers associated with persistent
names are quite different from identifiers associated with the
locations of objects. This is especially true when such names are
are expected to be stable for a very long time or when they identify
large and complex entities. In order to allow Uniform Resource Names
(URNs) to evolve to meet the needs of the Informational Sciences
community and other users, this specification separates the syntax
for URNs from the generic syntax for Uniform Resource Identifiers
(URIs) specified in RFC 3986, updating the latter specification
accordingly.
Status of this Memo
This Internet-Draft is submitted in full conformance with the
provisions of BCP 78 and BCP 79.
Internet-Drafts are working documents of the Internet Engineering
Task Force (IETF). Note that other groups may also distribute
working documents as Internet-Drafts. The list of current Internet-
Drafts is at http://datatracker.ietf.org/drafts/current/.
Internet-Drafts are draft documents valid for a maximum of six months
and may be updated, replaced, or obsoleted by other documents at any
time. It is inappropriate to use Internet-Drafts as reference
material or to cite them other than as "work in progress."
This Internet-Draft will expire on January 03, 2015.
Copyright Notice
Copyright (c) 2014 IETF Trust and the persons identified as the
document authors. All rights reserved.
This document is subject to BCP 78 and the IETF Trust's Legal
Provisions Relating to IETF Documents (http://trustee.ietf.org/
license-info) in effect on the date of publication of this document.
Please review these documents carefully, as they describe your rights
and restrictions with respect to this document. Code Components
extracted from this document must include Simplified BSD License text
as described in Section 4.e of the Trust Legal Provisions and are
provided without warranty as described in the Simplified BSD License.
Klensin Expires January 03, 2015 [Page 1]
Internet-Draft URNs are not URIs July 2014
Table of Contents
1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . 2
2. Pragmatic Goals . . . . . . . . . . . . . . . . . . . . . . . 3
3. A Perspective on Locations and Names . . . . . . . . . . . . . 3
4. Changes to RFC 3986 . . . . . . . . . . . . . . . . . . . . . 6
5. Other Required Actions . . . . . . . . . . . . . . . . . . . . 6
6. Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . 6
7. Contributors . . . . . . . . . . . . . . . . . . . . . . . . . 7
8. IANA Considerations . . . . . . . . . . . . . . . . . . . . . 7
9. Security Considerations . . . . . . . . . . . . . . . . . . . 7
10. References . . . . . . . . . . . . . . . . . . . . . . . . . . 7
10.1. Normative References . . . . . . . . . . . . . . . . . . 7
10.2. Informative References . . . . . . . . . . . . . . . . . 8
Appendix A. A More Pragmatic Perspective . . . . . . . . . . . . . 8
Appendix B. A Plausible Development Scenario . . . . . . . . . . . 9
Appendix C. Change Log . . . . . . . . . . . . . . . . . . . . . . 11
Appendix C.1. Changes from version -00 to -01 . . . . . . . . . 11
Author's Address . . . . . . . . . . . . . . . . . . . . . . . . . 11
1. Introduction
The Internet community now has many years of experience with both
name-type identifiers and location-based identifiers (or "references"
for those who are sensitive to the term "identifier" -- see Section
3). The primary examples of these two categories are Uniform
Resource Names (URNs [RFC2141] [RFC2141bis]) and Uniform Resource
Locators (URLs) [RFC1738]). That experience leads to the conclusion
that it is impractical to constrain URNs to the syntax and high-level
semantics of URLs. Generalization from URLs to generic Uniform
Resource Identifiers (URIs) [RFC3986], especially to name-based,
high-stability, long-persistence, identifiers such as many URNs, has
failed because the assumed similarities do not actually exist to a
sufficient degree. Ultimately, locators, which typically depend on
particular accessing protocols and a specification relative to some
physical space or network topology, are simply different creatures
from long-persistence, location-independent, object identifiers. The
syntax and semantic constraints that are appropriate for locators are
either irrelevant to or interfere with the needs of resource names as
a class. That was tolerable as long as the URN system didn't need
additional capabilities but experience since RFC 2141 was published
has shown that they are, in fact, needed.
Even then, it would have been possible to make URNs fit the "generic
URI" [RFC3986] bed by inventing a syntax with sufficient escapes and
embedding had the latter specification not also specified some
semantics for non-locational information. Whether such escapes and
embedding would have been a good idea is another matter: they tend to
make syntax more complex, harder for users to understand, and hence
Klensin Expires January 03, 2015 [Page 2]
Internet-Draft URNs are not URIs July 2014
more error-prone.
This specification updates the Generic URI Syntax specification
[RFC3986] to exclude URNs from its coverage. Put differently, with
the publication of this specification, URNs are no longer considered
a member of the class of URIs to which RFC 3986 applies.
[[Note in draft: the above leaves it ambiguous as to whether it
remains appropriate to call URNs "URIs". That ambiguity is
intentional and, if possible should keep the question part of the
"someone else's problem" category.]]
For URLs and such other URIs as may exist or be created in the
future, this specification does not change the syntax rules and other
requirements and recommendations of RFC 3986.
2. Pragmatic Goals
Despite the important background and rationale in the section that
follows, the change made by this specification is driven by a desire
to avoid philosophical debates about terminology or ultimate truths.
Instead, it is motivated by three very pragmatic principles:
1. Try to accommodate all of those who think URNs are necessary,
i.e., that they are distinct from URLs.
2. Try to avoid getting bogged down in declarative/ definitional
statements about what is and is not correct in the abstract.
3. Avoid a fork in the standard that leads to multiple, conflicting,
definitions or criteria for URNs.
3. A Perspective on Locations and Names
[[Note in Draft: See Appendix A for a different perspective.]]
Content industries (e.g., publishers) and memory organizations (e.g.,
libraries, archives, and museums) invest a lot of resources on naming
things and the topics of naming and classification are important
information science issues. Tens, if not hundreds, of millions of
persistent identifiers have been assigned during the last decade.
Several identifier systems have been developed for persistent and
unique identification of resources. When there is a real need to
preserve something important (such as scientific publications,
research data, government publications, etc.) for the long term, URNs
or other persistent identifiers are used; URLs (or other generic
URIs) are not being used for identification or even linking purposes.
Klensin Expires January 03, 2015 [Page 3]
Internet-Draft URNs are not URIs July 2014
Naming and locating, e.g., for library resources, are both complex
activities which have different aims. Traditionally, naming and
locating resources have been separate activities, and the rules for
the former are much more stringent than for the latter. The same
principles are being applied to digital materials as well as more
traditional ones. In a library, any book, be it printed or digital,
has both unique and persistent International Standard Book Number
(ISBN) and non-unique (each copy has its own location information)
and short-lived location information which cannot be trusted in the
long run. ISBN never changes, but both shelf locations and Web
addresses usually do, many times during the book's life span.
Giving location information a role in identification would not only
force libraries to adopt different policies for printed and digital
content, it would also undermine the value of existing identifier
systems. Let us assume that ten people independently upload a copy
of an electronic book into different locations in the Web. Are all
these ten URLs valid identifiers of the book? And what is their
relation to the ISBN or other identification information of the book
such as its title?
From the perspective of the communities who depend on persistent
identifiers, critical issues include:
1. Resource identification has to be a managed process. Assigning
URIs generally is not. Although it may be possible to introduce
some level of control to URI assignment, a user cannot determine
whether some URI is reliable or not.
2. Anyone may assign new URIs to resources even if these resources
already have proper identifiers assigned to them. Claiming that
these URIs actually identify something undermines the value of
proper identifiers.
3. There is no 1:1 relation between the resource identified and
URIs. An e-book in the Web may be represented as 1-n files
(URIs), and a single file may contain several books. And books
are simple, we need to name very complex objects such as research
data sets, or some component parts within these complex data
sets.
4. One resource such as a scientific article is typically available
from multiple locations, including (for instance) the publisher's
document supply service, a university's open repositories and
other cooperative repository systems, legal deposit collections
and the Internet archive. A resource should have one and only
one identifier of a given type; URIs do not meet this
requirement.
Klensin Expires January 03, 2015 [Page 4]
Internet-Draft URNs are not URIs July 2014
5. URIs relate to instances (copies) of resources, whereas
traditionally identification has much broader scope. Identifiers
may be assigned to, e.g., an immaterial work (such as Hamlet),
its expressions (e.g. Finnish translation of Hamlet), and
manifestations of works and expressions (e.g. PDF version of
Finnish translation of Hamlet).
6. Over time, different resources (or different versions of the same
resource) may be found from the same non-URN URI. A user has no
way of knowing whether the resource has changed. One of the
basic principles for proper identifier systems is that the same
identifier is never assigned to another resource. In general,
URIs do not meet this requirement.
7. Persistent identification must be available for resources which
are available only in databases and other environments that are
often identified today as "deep web". URIs for these resources
tend to be very complicated and it will be difficult to keep them
alive even with the help of DNS redirection when e.g. the
underlying database management system changes.
8. The role URI fragment and query could or should have in
identification is unclear and the statements in RFC 3986 are
definitely problematic from the points of view of existing
identifier systems and management of naming.
Does "fragment" identify a location or a certain section of a
resource? In the evolving set of URN Internet standards,
fragment will not be a part of the Namespace Specific String.
Then fragment only indicates a place / segment within the
identified resource, but does not identify it. If fragment
had a role in identification, fragments would extend the scope
of existing standard identifiers to component parts of
resources. For instance, anyone could use URN based on ISBN +
fragment to identify chapters of electronic books.
Things get even more complicated with "query" since what the
combination of an identifier and a query resolves to may not
have anything to do with the original resource. For instance,
a URN based in ISBN + query may resolve to the metadata record
describing the book. These records have their own identifiers
which are not based on ISBNs.
[[Note in draft: Most of the discussion above may belong in
2141bis and/or 3406bis rather than here.]]
9. For many organizations, persistence means decades or centuries.
Anything that is protocol dependent will eventually fail. URLs
do not change by themselves, but in the long run it is very
difficult for people to not change them or the objects to which
they point.
Klensin Expires January 03, 2015 [Page 5]
Internet-Draft URNs are not URIs July 2014
The mention of centuries is intentional. Content industries,
memory organizations (such as national and repository
libraries and national archives) and universities and other
research organizations, need identifiers that will persist for
hundreds of years. Such identifiers might even need to
outlast the institutions themselves, and definitely should be
usable even if current technologies such as the Web and the
Internet cease to exist or are supplanted by something new (as
unlikely as that might seem today).
In addition, operations on, or additional specifications
about, names and the associated objects must be possible, as
stable as the names themselves, and reasonably efficient. For
example, if a URN were assigned to an encyclopedia that
consisted of many volumes, it should be feasible to identify
(and locate and retrieve if that were desired) a particular
volume or even a particular article without accessing or
retrieving the entire set.
4. Changes to RFC 3986
This specification removes URNs from the scope of RFC 3896. It makes
no changes for URI types that remain within that scope and has no
practical effect for URNs defined in strict conformance to the prior
URN specification [RFC2141] or the associated registration
specification [RFC3406].
5. Other Required Actions
The basic URN syntax specification [RFC2141] was published well
before RFC 3986 and therefore does not depend on it. Successors to
that specification will need to fully spell out the syntax and
semantics of URNs, eliminating or using great care about generic or
implicit reference to any URI specification.
6. Acknowledgments
Klensin Expires January 03, 2015 [Page 6]
Internet-Draft URNs are not URIs July 2014
This specification was inspired by a search in the IETF URNBIS WG for
other alternatives that would both satisfy the needs of persistent
name-type identifiers and still fully conform to the specifications
and intent of RFC 3986. That search lasted several years and
considered many alternatives. Discussions with Leslie Daigle, Juha
Hakala, Barry Leiba, Keith Moore, Andrew Newton, and Peter Saint-
Andre during the last quarter of 2013 and the first quarter of 2014
were particularly helpful in getting to the conclusion that a
conceptual separation of notions of location-based identifiers (e.g.,
URLs) and the types of persistent identifiers represented by URNs was
necessary. As noted below, Juha Hakala provided much of the text on
which Section 3 was based. Peter Saint-Andre provided significant
text in a pre-publication review. The author also appreciates the
efforts of several people, notably Tim Berners-Lee, Julian Reschke,
Lars Svensson, Henry S. Thompson, and Dale Worely, to challenge text
and ideas and demand answers to hard questions. Whether they agree
with the results or not, their insights have contributed
significantly to whatever clarity and precision appears in the text.
7. Contributors
Juha Hakala contributed most of the text of Section 3.
Contact Information:
Juha Hakala
The National Library of Finland
P.O. Box 15, Helsinki University
Helsinki, MA FIN-00014
Finland
Email: juha.hakala@helsinki.fi
8. IANA Considerations
[[RFC Editor: Please remove this section before publication.]]
This memo is not believed to require any action on IANA's part. In
particular, we note that there are a collection of "Uniform Resource
Identifier (URI) Schemes" that does not include URNs and a series of
URN-specific registries that do not rely on the URI specificstions.
9. Security Considerations
This specification changes the structural syntax and semantics of
URNs to make them self-contained (as specified in other documents)
rather than making them dependent on generic URI syntax. It should
have no effect on Internet security unless the use of a definition
and syntax that are more clear reduces the potential for confusion
and consequent vulnerabilities.
10. References
10.1. Normative References
Klensin Expires January 03, 2015 [Page 7]
Internet-Draft URNs are not URIs July 2014
[RFC2141] Moats, R., "URN Syntax", RFC 2141, May 1997.
[RFC3986] Berners-Lee, T., Fielding, R. and L. Masinter, "Uniform
Resource Identifier (URI): Generic Syntax", STD 66, RFC
3986, January 2005.
10.2. Informative References
[DeterministicURI]
Mazahir, O., Thaler, D. and G. Montenegro, "Deterministic
URI Encoding", February 2014, <http://www.ietf.org/id/
draft-montenegro-httpbis-uri-encoding-00.txt>.
[RFC1738] Berners-Lee, T., Masinter, L. and M. McCahill, "Uniform
Resource Locators (URL)", RFC 1738, December 1994.
[RFC2141bis]
Saint-Andre, P., "Uniform Resource Name (URN) Syntax",
January 2014, <https://datatracker.ietf.org/doc/draft-
ietf-urnbis-rfc2141bis-urn/>.
[RFC3406] Daigle, L., van Gulik, D., Iannella, R. and P. Faltstrom,
"Uniform Resource Names (URN) Namespace Definition
Mechanisms", BCP 66, RFC 3406, October 2002.
Appendix A. A More Pragmatic Perspective
[[The community should decide whether this appendix, or a modified
version of it, should remain or be removed at the time of RFC
publication. In principle, it could even be retained by splitting
the relevant Section above into two parts and making a variation on
the text below into one of them. Those who think it should be
retained are encouraged to supply text.]]
Section 3 provides an explanation of the reasons for this change.
That explanation is not without controversy, especially from those
who make different assumptions about the future, or even
interpretations of the present, than many members of the community
(and especially members of the communities described in that
section). Some of those who do not accept the explanation above
simply do not recognize the distinctions on which it, and URNs more
generally, are based, including the name-locator distinction. In
some cases, opposition to that explanation is quite pronounced,
Klensin Expires January 03, 2015 [Page 8]
Internet-Draft URNs are not URIs July 2014
involving fundamental differences in philosophy that move beyond mere
differences of opinion.
Like most controversies in which one group does not accept the
definitions, facts, or logic of another, the differences are unlikely
to be resolved by further discussion, no matter how sensible and
patient. The material in this appendix is provided for the benefit
of those who cannot accept Section 3 or consider the discussion there
to be meaningless.
Independent of the details of the discussion above, in the case of
URNs, the IETF is faced with a pair of problems that are ultimately
faced sooner or later by all voluntary standards bodies: nothing
except quality and broad community consensus prevents a standard from
being ignored in the marketplace and nothing prevents another body
from creating a competing standard. The effort required to create a
competing standard can be increased and its potential for confusion
can be reduced somewhat by various measures -- measures the IETF has
rarely tried to actually use -- but those measures are rarely
effective when the other body is convinced that they have legitimate
and significant needs that differ from the original atandard.
Because of those problems, the key question for the URN effort is
ultimately not whether a clear enough distinction exists between
names and locator or location-based information, nor whether
"persistent" can be defined clearly enough, nor even whether the
communities and requirements described in Section 3 are valid or will
be judged valid in retrospect in a few decades or centuries.
Instead, the question is whether the IETF is willing to evolve and
adapt the URN definition to accommodate those perceived needs or
whether if prefers to have that work done elsewhere, either by
adoption in the broader community and marketplace of a different
approach or, potentially, even a competing URN standard. If, in the
long run, those other communities and perspectives turn out to be
wrong, the additional features will atrophy. But that would be true
whether they are specified and standardized in the IETF or elsewhere.
Appendix B. A Plausible Development Scenario
NOTE IN DRAFT: this appendix is included in draft -01 to summarize
some discussions on the mailing list in May and June 2014 for the
convenience of the WG and possible discussions at IETF 90. It really
is not part of this document and will be removed in the next version.
The question has come up several times about what a URN syntax might
look like when the URI (i.e., RFC 3986) constraints were removed and
the questions about matching and resolution mechanisms that have
plagued the WG were addressed.
Klensin Expires January 03, 2015 [Page 9]
Internet-Draft URNs are not URIs July 2014
One possible answer is that, if some of those questions can be
successfully ignored, e.g., by never having the equivalent of query
or fragment components treated as part of the URN for matching
purposes and by using the same resolution framework for all URN, one
could preserve the generic URI syntax, effectively just using this
specification to break the link with some of the semantics specified
in RFC 3986. That strategy makes sense only if the IETF is convinced
that it understands all present and potential URNs well enough to
specify those properties globally, possibly using an IANA registry
for pointers to resolution mechanisms (at one per URN NID) for which
see below.
If the community is less confident that it understands the full range
of requirements for future URN namespaces, then one might, for
example, generalize URNs and extend RFC 2141 so that a URN was,
conceptually,
"URN" NID NSS [ServiceRequests...]
It will ultimately make a difference whether "a URN" is the complete
URN string as above or just urn:NID:NSS or, put differently, whether
the ServiceRequests are part of the NSS. But it makes less
difference in the near term than out trying to make general URIs work
for URNs would suggest.
In the above, a ServiceRequest is, again conceptually, a tuple of
ServiceType ComparisonIndicator ServiceTarget RequestParameters...
ServiceType is nominally some sort of keyword. ComparisonIndicator
tells something trying to compare a pair of URNs for identity whether
that particular ServiceRequest counts or should be ignored.
ServiceTarget identifies where the ServiceRequest is to be sent and,
depending on the ServiceType, may be a keyword indicator or, at the
risk of descending into recursion hell, a URL or URN. And
RequestParameters are anything the ServiceType definition says they
are.
Any of those may be
o defined in the NID registration and omitted from (prohibited in)
the URN string
o allowed by the NID registration but explicitly included in the URN
string
o defined in the NID registration as a default but allowed in the
URN string as an override
o prohibited entirely by the NID registration (effectively
duplicating the "don't do that" rule of 2141 on a per-NID basis).
Klensin Expires January 03, 2015 [Page 10]
Internet-Draft URNs are not URIs July 2014
In addition, ServiceTarget might be specified in the NID registration
to identify an IANA registry or domain subtree.
Presumably the NID registration may also specify whether anything not
required is prohibited, and the other variations on that theme.
Requests/specifications for location information, assorted metadata,
or model or actual objects themselves are then just specialized
ServiceRequests. In particular, "Fragment" disappears as a special
type of syntax and reappears as a Service Request that is applicable
to some NIDs and not others and whose meaning and action (and how it
is "resolved") are specified on an NID basis and as above. "Query"
disappears too, not because it (or the syntax) are necessarily
problematic but because the term itself is misleading for many
possible types of ServiceRequests and therefore causes more confusion
than it clears up.
ALmost independent of the above, unless we globally allow or prohibit
non-ASCII content in URN strings, the registration/ definition of the
NID would presumably identify what characters are permitted and, if
necessary, how they are interpreted for matching purposes.
Appendix C. Change Log
RFC Editor: Please remove this appendix before publication.
Appendix C.1. Changes from version -00 to -01
o Revised Section 1 slightly and added some new material to try to
address questions raised on the mailing list.
o Added Section 2, reflecting an email exchange.
o Added a Security Considerations section, replacing the placeholder
in the previous version.
o Added Appendix Appendix A and inserted a note in Section 3
pointing to it.
o Added temporary Appendix Appendix B for this version only.
o Enhanced and updated the Acknowledgments section.
o The usual small clarifications and editorial changes.
Author's Address
Klensin Expires January 03, 2015 [Page 11]
Internet-Draft URNs are not URIs July 2014
John C Klensin
1770 Massachusetts Ave, Ste 322
Cambridge, MA 02140
USA
Phone: +1 617 245 1457
Email: john-ietf@jck.com
Klensin Expires January 03, 2015 [Page 12]
