Internet Engineering Task Force Ron Daniel Jr. INTERNET-DRAFT Los Alamos National Laboratory draft-ietf-uri-urc-sgml-00.txt Terry Allen O'Reilly and Associates, Inc. June 16, 1995 An SGML-based URC Service Status of this draft This document is an Internet-Draft. Internet-Drafts are working documents of the Internet Engineering Task Force (IETF), its areas, and its working groups. Note that other groups may also distribute working documents as Internet-Drafts. Internet-Drafts are draft documents valid for a maximum of six months. Internet-Drafts may be updated, replaced, or obsoleted by other documents at any time. It is not appropriate to use Internet-Drafts as reference material or to cite them other than as a ``working draft'' or ``work in progress.'' To learn the current status of any Internet-Draft, please check the 1id-abstracts.txt listing contained in the Internet-Drafts Shadow Directories on ds.internic.net, nic.nordu.net, ftp.isi.edu, or munnari.oz.au. This Internet Draft expires December 15, 1995. Abstract The URI Working Group has been developing an architecture where Internet resources are identified using a Uniform Resource Name (URN), and retrieved using a Uniform Resource Locator (URL). Mapping URNs to URLs is the job of the Uniform Resource Characteristics (URC) service, whose requirements were given in [1]. This paper presents one possible specification for the URC service. This spec provides the means for the URC service to formally specify new capabilities, while retaining the speed that is paramount to the fundamental use of the URC service as the means for URN to URL resolution. INTERNET-DRAFT An SGML-based URC Service June 16, 1995 Contents 1 Introduction 4 2 URN Resolution Overview 5 3 Attribute Sets 5 4 Default Attribute Set 8 5 Multiple Syntaxes 10 5.1 Example 1: text/html . . . . . . . . . . . . . . . . . . . . 10 5.2 Example 2: text/urc0 . . . . . . . . . . . . . . . . . . . . 12 5.3 Example 3: text/sgml . . . . . . . . . . . . . . . . . . . . 12 6 Query Languages 13 6.1 Trivial Query Language . . . . . . . . . . . . . . . . . . . 14 6.2 Query Language Identification . . . . . . . . . . . . . . . . 15 7 Requirements Satisfaction 16 7.1 Requirements on the URC . . . . . . . . . . . . . . . . . . . 16 7.2 Requirements on the URC Service . . . . . . . . . . . . . . . 18 8 Random Notes 19 8.1 Other IETF Metadata Efforts . . . . . . . . . . . . . . . . . 19 8.2 Referring to Particular URCs . . . . . . . . . . . . . . . . 20 9 Open Issues 20 9.1 Query Language . . . . . . . . . . . . . . . . . . . . . . . 20 9.2 Meta-metadata . . . . . . . . . . . . . . . . . . . . . . . . 20 9.3 Attribute set object encapsulation . . . . . . . . . . . . . 21 10Acknowledgments 21 draft-ietf-uri-urc-sgml-00 [Page 2] INTERNET-DRAFT An SGML-based URC Service June 16, 1995 11References 22 12Security Considerations 22 A SGML Declaration 23 B DTD for Default Attribute Set 25 C DTD for Meta Attribute Set 30 D Custom Attribute Set Example 31 draft-ietf-uri-urc-sgml-00 [Page 3] INTERNET-DRAFT An SGML-based URC Service June 16, 1995 1 Introduction Experience with the WWW has exposed the problems inherent in basing the system on resource locations instead of resource identity. Uniform Resource Locators (URLs) typically identify a particular path on a particular host. This leads to a bevy of problems with network hotspots, fault-tolerance, and resource management. To overcome those problems, the Uniform Resource Identifiers Working Group of the Internet Engineering Task Force has been developing an architecture that uses Uniform Resource Names (URNs) for resource identification. A name resolution service would handle the problem of mapping names to locations for the purpose of retrieval. The data structures that contain the information necessary for this resolution are known as Uniform Resource Characteristics (URCs), and the resolution service is known as the URC service. Several scenarios of how this service would be used, and the requirements they place on the service, were set forth in [1]. The primary purpose of the URC service is to resolve URNs to URLs. However, the URC makes too good a place to store additional information about the resource to pass up the opportunity. It is easy to imagine storing basic bibliographic information, such as author, title, and subject, in order to provide the foundation for a "card catalog" service for Internet-accessible documents. Of course, there is no reason to stop with documents. Scientific datasets, product databases, computer-generated music, etc. are all reasonable candidates for publication over the WWW. The more one looks at the URC service, the more one realizes just how great a range of information it could reasonably provide. This leads us to looking at the URC service as a general service for presenting metadata - or data about data. Because of the wide variety of data that can be made available over the Internet, and because of the diversity of the metadata we might want to use to describe it, no single set of attributes (such as author, title, subject) are universally applicable. This argues for a very general means of specifying attribute sets. At the same time, recall that the primary purpose of the URC service is for URN to URL resolution. This argues for a single, easily parsed, attribute set. Other apparently conflicting requirements were set forth in [1]. This proposed specification attempts to reconcile these conflicting demands. The need for a formal method is met by using SGML Document Type Definitions (DTDs) to specify the structure of new attribute sets. This is described in section 3. Simple changes to attribute sets can be accommodated through a single-inheritance mechanisms that is also described in section 3. The need for fast, heuristic parsing is met by providing a particular DTD that is believed to be widely, though not universally, applicable. The resolution process is reviewed in section 2. The default attribute set is described in section 4 and supplied in Appendix B. The specification allows for user agents to request URC information in different transfer syntaxes draft-ietf-uri-urc-sgml-00 [Page 4] INTERNET-DRAFT An SGML-based URC Service June 16, 1995 in order to ease parsing or provide particular capabilities, such as digital signatures. The ability for using multiple syntaxes is described in section 5. Another important part of the service is the means that it provides for queries. The specification allows for multiple query languages. This part of the spec is described in section 6, which also describes the trivial query language that all URC servers must support. How the specification meets the requirements established in [1] is the subject for section 7. Section 8 contains random notes and section 9 discusses issues that are still unresolved at this time. This is the first draft of the specification, and it is known to be incomplete. It makes no attempt to discuss how URC information will be stored at a server, and does not address issues of maintaining URCs, distributing the database for fault-tolerance, etc. 2 URN Resolution Overview A variety of URN syntaxes and resolution procedures are being studied by the URI-WG. This spec assumes a syntax and resolution procedure roughly like that in [4]. Briefly, such a URN contains a Fully Qualified Domain Name (FQDN), which identifies a set of servers that are authorized by the publisher to resolve the publisher's URNs. (These are known as default URNs). The client sends an HTTP GET request for the complete URN to that server. The request may use HTTP's Accept: header to indicate preferences for the results to be returned in particular syntaxes. The result is returned to the browser. Depending on the transfer syntax and browser capabilities, the browser may choose one of several URLs itself, it may hand the URC off to an external application that can make the selection, or the browser may display the URC to the user so the user can make the selection. This specification uses HTTP as the resolution protocol. Use is made of HTTP's format negotiation capabilities. Using HTTP should ease the transition to more secure resolvers, which is a requirement, because of S-HTTP, SSL, and similar security efforts. Furthermore, a wide variety of browsers, servers, tools, and expertise already exist for HTTP and can quickly be brought to bear on the URC service. 3 Attribute Sets The primary purpose of the URC service is to resolve URNs into URLs for the purpose of resource retrieval. However, the URC makes a very convenient place to store metadata - data about the resource. Frequently this will be bibliographic information, but [1] requires draft-ietf-uri-urc-sgml-00 [Page 5] INTERNET-DRAFT An SGML-based URC Service June 16, 1995 that there be no restrictions on the data that can be placed in the URC. The URC is intended to be a container for metadata about a wide variety of Internet resources. Satellite images, poems, scientific datasets, fine art images, gene sequences, ... are all reasonable candidates for publication in the Internet's Uniform Resource Architecture. All of these resource types will need different sorts of metadata. Other attributes, such as ``subject'', may be used in different fashions. Because of this diversity, we make the fundamental assumption: There are no metadata elements (such as author, title, subject, etc.) that are applicable to all resources. Because of this assumption, we need a means of specifying what attributes are being used in a particular URC, as well as their syntax and semantics. This need brings up the notion of an attribute set and the attribute set identifier. Definitions: An Attribute Set (AS) is the particular collection of elements that may appear in a particular URC. An Attribute Set Definition (ASD) is a machine-parsable specification of the elements in an attribute set, and the grammar by which they may be combined. An Attribute Set Identifier (AID) is a URN that can be resolved to obtain the attribute set definition. Using a URN to identify the attribute set of a URC has two advantages. First, URNs are unambiguous, so we can tell if the contents of one ``subject'' field are comparable to another. Second, using a URN lets us retrieve the attribute set definition if we need to. The definition is a machine parsable grammar specification for the URCs. This allows us to parse novel URCs, although dealing with the semantics of novel elements is still an unsolved problem. A further enhancement to this model is that an AS can be a modification of an existing AS. The child AS would specify only the additions and changes to the parent AS. Thus, attribute sets can form a single inheritance scheme back to some, presumably well-known, base attribute set. Multiple Inheritance (MI) of attribute sets was considered and explicitly rejected for reasons of complexity, robustness, complexity, poor behavior in distributed systems, complexity, lack of universal language support, and complexity. Furthermore, at least one of the authors (RD) believes that MI is just too complex. Dig? The attribute set definition shall be an SGML DTD. Parameter entities shall be used to allow element definitions to be overridden in a single inheritance scheme. Such an approach is illustrated in draft-ietf-uri-urc-sgml-00 [Page 6] INTERNET-DRAFT An SGML-based URC Service June 16, 1995 Appendices B, C, and D. The AS definition specifies the syntax of a URC in a machine-usable fashion. There are three complications to this model. First, we must also provide a specification of the semantics of the elements. At this time, we are unaware of any machine-usable semantic specification schemes with the generality needed for the URC task. Therefore, we rely on human-readable specification of the semantics of the elements. The semantics of the elements in the attribute set shall be indicated by comments in the DTD. Check w/ comp.text.sgml types on schemes for automatically extracting comments for documentation purposes. Another mechanism is available for locating machine-parsable semantic definitions once they become available. But before we can describe that, we must talk about the other complications. A second complication concerns the URC syntax. Defining the attribute set with an SGML DTD only allows us to automatically parse URCs that are conveyed in an SGML transfer syntax. Note that other syntaxes are explicitly allowed as a feature of this proposal. Thus, if a request is made for a text/plain syntax, the result is not parsable using the AS definition. This is not a great problem. First, it is easy enough to request the URC in a text/sgml syntax, which is required to be conformant with the AS DTD. Second, we rarely care about parsing according to ISO 8879. Because the primary use for the URC service is URN to URL resolution, we will usually parse the URC in a heuristic fashion, rather than spend the time to retrieve all the inherited DTD fragments, build the DTD, parse the document according to it, and then find that we have a lot of elements whose meaning we don't know anyway. The default AS is provided to simplify the task of heuristic parsing. In fact, the trivial query language provides the means for the client to ask for just a single location of the resource. More capable query languages will offer the ability to ask for arbitrary subsets of the URC information. A third complication arises as a result of using a URN for the AID. Assume we have retrieved a URC, call it URC-1, that specifies its AID (AID-1). Also assume that we wish to retrieve the attribute set definition. We resolve AID-1, which is a URN, and get back a URC (URC-2) that lists locations for the AS definition. What is the AID in URC-2? How do we avoid infinite regress? This standard defines a basic meta-attribute set definition that is suitable for the URC of an attribute set (see Appendix C). To avoid infinite regress, AIDs can either be a URN, or the distinguished string "root". Providing a URC for the AS definition is a complication, but it also provides us with a natural extension mechanism for dealing with the semantics of an attribute set. Just as a normal document might have ASCII and PostScript representations, the AS definition might have SGML and KQML representations. These alternate representations are how we can provide versions of an AS definition with machine-readable semantic definitions. draft-ietf-uri-urc-sgml-00 [Page 7] INTERNET-DRAFT An SGML-based URC Service June 16, 1995 4 Default Attribute Set The use of inherited attribute set grammars provides a very general capability, and there are applications that can use that generality to improve their functionality and scalability. However, the primary purpose of the URC service is URN to URL resolution, and the speed of that resolution is a major concern for the URC service. Response times for interactive browsing do not allow multiple network accesses in order to fetch DTD fragments and build a grammar for parsing a URC. Therefore, we provide a default attribute set whose semantics are to be broadly understood. If a URC comes in that has been described using this attribute set, then it can be parsed in either a formal (according to ISO 8879) or a heuristic (anything else) fashion. If a URC comes in with no attribute set specified, then the default attribute set is assumed. Furthermore, if a URC comes in with a different attribute set, we mandate that any elements it has with the same names as elements in the default attribute set must have similar semantics to the elements of the default set. This will allow heuristic parsing of new attribute sets. To amplify this last point, consider the TITLE element. Anyone creating a new attribute set that contains TITLE must use it in a fashion similar to that in the default attribute set. If someone wants to use TITLE for a very different purpose (describing royalty perhaps), they must add special information (scheme, type, or other attributes) so that it is possible to tell that TITLE is being used in a novel fashion and simple parsers are not led astray. For the purposes of this draft of the specification, and the prototypes developed from it, the AID for the default attribute set shall be: It is anticipated that as this specification moves to the standards track, IANA will provide the URN for the default attribute set. The default attribute set is strongly based on the work of the Dublin metadata workshop [8]. A brief overview is given below. The attribute set is based on a few principles. First, we provide a set of elements with widely understood semantics (such as ). Second, we provide a mechanism for specifying more precise interpretations of those elements (such as <Subject scheme="LCSH">). Third, we allow a variety of transfer syntaxes for the attributes. Plain text, HTML, binary encodings, etc. can all be used, as described further in section 5. All the elements in the default attribute set are optional and repeatable. No particular order is required. Identifier: String or number used to uniquely identify this object. draft-ietf-uri-urc-sgml-00 [Page 8] INTERNET-DRAFT An SGML-based URC Service June 16, 1995 Normally a URN will be the unique identifier, and the <URN> element will be used instead. URN: Shorthand for <Identifier scheme="URN">. Instance: Instance is a construct for grouping information on particular instances of a resource, such as location, format, price, etc. Location: A string to represent information on how to retrive the resource from a particular location. Usually the <URL> element will be used instead, but this element is here for things like LIFNs, etc. URL: Shorthand for <Location scheme="URL">. The content is a URL to fetch an instance of the resource. Author: The person(s) and/or organization(s) primarily responsible for the intellectual content of the work. Title: The name of the object. Subject: The field of knowledge to which the work belongs. Publisher: The agent or agency responsible for making the object available. Date: The date of publication. Other Agent: Other person(s) and/or organization(s), such as editors, transcribers, sponsors, etc. who have made significant contributions to the work. Author and Publisher are special cases of OtherAgent. Object type: The abstract category of the object, such as novel, poem, dictionary. Form: The particular manifestation or data representation of the object, such as PostScript file or Windows executable. For URCs, form will typically be specified as an Internet Media Type - formerly known as the MIME Content-type. Relation: Relationship to other objects. This element should identify the role of the relationship, as well as the related objects. Source: Objects, either print or electronic, from which this resource was derived. This is a special case of the Relation element that is believed to be widely useful to the humanities. Language: Natural language of the intellectual content. draft-ietf-uri-urc-sgml-00 [Page 9] INTERNET-DRAFT An SGML-based URC Service June 16, 1995 Coverage: The spatial locations and temporal durations characteristic of the object. 5 Multiple Syntaxes Several of the URC requirements are difficult, if not impossible, to satisfy at the same time. For example, it is a requirement to have a human-readable, printed representation of a URC. It is also a requirement that URCs have a consistent encoding that is suitable for digital signature computation. Unfortunately, end-of-line convention differences between platforms make it difficult to meet both requirements simultaneously. A related problem is that there will be different uses for URC information, and different encodings will be appropriate to meet those different needs. For example, it might be useful to obtain citations in some format (BiBTeX perhaps) for inclusion in another system. Because of these considerations, we make another fundamental assumption: There is no universally applicable syntax for a URC. Because of this assumption, this specification calls for the use of format negotiation mechanisms, specifically the Accept: header in HTTP [2], to indicate preferred syntaxes. Depending on the user's intent for the URC, they can ask for an HTML encoding, a plain text encoding, an encrypted binary encoding, a synthesized audio rendition, etc. A set of examples are provided that show the requests and responses for different renditions of the same underlying information. All of these examples assume we are resolving the URN: urn:dns:pchs.k-12.okc.ok.us:student-papers-1995/geo3 which is a scandalous story written by a student at my old high school. 5.1 Example 1: text/html When the URC service is first deployed, there will be a large base of existing web browsers that will not have the ability to parse URCs. draft-ietf-uri-urc-sgml-00 [Page 10] INTERNET-DRAFT An SGML-based URC Service June 16, 1995 One means of remaining compatible with these browsers is to request the URC in HTML. The user can look at information on the different locations for a resource, pick a site, and click on a link to retrieve a resource. The URN resolution request sent to the URC server's HTTP daemon might be: GET student-papers-1995/geo3 HTTP/1.0 Accept: text/html, */*; q=0.2 which says to send text/html if possible, anything else if not. The reply from the server might be: HTTP/1.0 200 OK Date: Tuesday, 08-Oct-96 21:09:16 GMT Server: Apache/2.7 MIME-version: 1.0 Content-type: text/html Last-modified: Friday, 26-Apr-96 21:57:12 GMT Content-length: 829 <html> <head> <title>URC for urn:x-dns-2:pchs.k-12.okc.ok.us:student-papers-1995/geo3

urn:x-dns-2:pchs.k-12.okc.ok.us:student-papers-1995/geo3

Author: Smith, Fred
Title: A Vicious, Seditious, and Tendentious History of George III
Subject: American Revolution Subject: (In)famous crackpots of history Location: http://www.pchs.k-12.okc.ok.us/student-papers/1995/smith/geo3.html
Form:text/html
draft-ietf-uri-urc-sgml-00 [Page 11] INTERNET-DRAFT An SGML-based URC Service June 16, 1995 5.2 Example 2: text/urc0 The text/html example above provides one means for backward compatibility with legacy clients. However, the "click twice" model is going to get old really quickly. A CCI helper application could be constructed that would parse URCs in some trivial format, such as text/urc0 [7], and automatically pick a URL. GET student-papers-1995/geo3 HTTP/1.0 Accept: text/urc0, */*; q=0.2 HTTP/1.0 200 OK Date: Tuesday, 08-Oct-96 21:09:16 GMT Server: Apache/2.7 MIME-version: 1.0 Content-type: text/urc0 Last-modified: Friday, 26-Apr-96 21:57:12 GMT Content-length: 62 ===== http://www.pchs.k-12.okc.ok.us/student-papers/1995/smith/geo3.html 5.3 Example 3: text/sgml This example assumes the URC is prepared according to the default attribute set. Note that the document declaration is included in the response, and that a URN is used in the SYSTEM identifier. This is how the attribute set is indicated for the SGML syntax. This is somewhat contrary to the intent of SGML, which wants to use SYSTEM to locate local information, and PUBLIC for well-known, publicly accessible information. I have gone against that convention, since SGML PUBLIC identifiers have a restricted character set, perform some processing on the characters, and encourage a syntax (known as a formal public identifier) that has not yet been mapped to onto URNs. By using a SYSTEM identifier we get a more liberal character set, and the string is handed off unmolested to an ``entity manager'' for processing so as to fetch the appropriate file. Using a URN here will require an enhanced entity manager, but such things are already part of at least one commercial SGML/WWW product. The URN resolution request sent to the URC server's HTTP daemon might be: GET student-papers-1995/geo3 HTTP/1.0 draft-ietf-uri-urc-sgml-00 [Page 12] INTERNET-DRAFT An SGML-based URC Service June 16, 1995 Accept: text/sgml, */*; q=0.2 The reply from the server might be: HTTP/1.0 200 OK Date: Tuesday, 08-Oct-96 21:09:16 GMT Server: Apache/2.7 MIME-version: 1.0 Content-type: text/sgml Last-modified: Friday, 26-Apr-96 21:57:12 GMT Content-length: 129 urn:dns:pchs.k-12.okc.ok.us:student-papers-1995/geo3 Smith, Fred A Vicious, Seditious, and Tendentious History of George III American Revolution (In)famous crackpots of history http://www.pchs.k-12.okc.ok.us/student-papers/1995/smith/geo3.html
text/html
 The examples above have suggested particular syntaxes. Determining the set of ``well-known'' syntaxes that all URC servers must be able to emit is one direction this standard could be extended. 6 Query Languages Diversity is the hallmark of the URC service. We want to encourage the formation of a variety of value-added services, therefore we need a standard means of dealing with unique capabilities. For example, different search services will have different information and support different forms of queries. This section defines a trivial query draft-ietf-uri-urc-sgml-00 [Page 13] INTERNET-DRAFT An SGML-based URC Service June 16, 1995 language that all URC servers and clients must support in order to claim conformance with the standard. This section also describes the scheme by which non-standard queries can be identified. 6.1 Trivial Query Language This query language is based on work in [3] and [4]. Queries in the trivial query language are HTTP GET and POST requests. The following grammar describes the trivial query language. It assumes a URN syntax similar to that in [4]. query ::== [N2operation/]urn | L2operation/url | C2operation/urc | Q-id/string N2operation ::== N2L | N2C L2operation ::== L2C C2operation ::== C2C Q-id ::== "Q-" string urn ::== name`space:FQDN[:port]:string url ::== scheme:scheme-specific-string urc ::== some`hairy`string The query operations are: N2L: Given a URN, return a URL. Typically the URL will be returned as an HTTP Location: header, so that clients will automatically retrive the resource. N2C: Given a URN, return a URC. The syntax for the result is negotiated by use of the HTTP Accept: header. This operation is the default, so that proxies may cache the result. L2C: Given a URL, return all the URCs that contain it. C2C: Given a partial URC, return all the URCs that match it. This part of the specification is being developed and should be the subject of a revised Internet-draft. Q-id: An extension mechanism for site-specific queries. draft-ietf-uri-urc-sgml-00 [Page 14] INTERNET-DRAFT An SGML-based URC Service June 16, 1995 The N2 and L2 operations are sent as HTTP GET requests. The C2C operation is sent as an HTTP POST request since it is likely to be quite large. The trivial query language also reserves a section of the URN namespace to obtain information on the resolver, resources, etc. They all begin with the reserved string "urn+". The requests and their meaning are: o urn+m Resolver meta-information (e.g. GET N2C/urn:x-dns-2:FQDN:urn+m) Returns a URC indicating URLs where resolver metainformation, such as the administrative contact, sponsoring organization, publication policy, etc. can be retrieved. o urn+a list of All RequestIDs (e.g. ) Returns a URC indicating URLs where a list of all URNs provided by the publisher can be obtained. The request must be understood by all resolvers, but the response may have zero URLs in it. In such a case the URC should contain a message saying the equivalent of ``sorry''. o urn+c Child naming authorities (e.g. ) Returns a URC indicating URLs where a list of any child naming authorities licensed by the naming authority associated with the FQDN can be found. o urn+p Parent naming authorities (e.g. ) Returns a URC indicating URLs where a list of any parent naming authorities licensing the naming authority associated with the FQDN can be found. o urn+aids (e.g. ) Returns a URC indicating the URLs that can be used to fetch a list of all the attribute sets used to describe information on this server. 6.2 Query Language Identification This relatively minimal set of queries does not allow the construction of complex queries, such as "gimme the URNs and titles of all resources with 'Smith' as an author and 'food' as the subject". Different organizations will wish to add value to their URC collections in different fashions, so it is highly unlikely that one draft-ietf-uri-urc-sgml-00 [Page 15] INTERNET-DRAFT An SGML-based URC Service June 16, 1995 query language will meet all needs. Therefore, we provide an additional well-known query operation, Q-id, given in the grammar above. Sites provide a unique identifier for the ID string, a slash, then a query string in their choice of query languages. It is suggested that the query ID string be a URN that will resolve to a description of the query language, its grammar, etc. On a related note, an additional URN is reserved: o urn+qls (e.g. ) Returns a URC that gives URNs and URLs for all the query languages known by this server, their conditions of use, etc. 7 Requirements Satisfaction This section reviews how the proposed URC specification meets the requirements set forth in [1]. That document divided the requirements into two categories: requirements on the URC and requirements on the URC service. 7.1 Requirements on the URC Machine Consumption Consistent External Representation Transport Friendliness Readability: The scenarios paper set forth several requirements on the URC and its encoding. The first set of those requirements were Machine Consumption, Consistent External Representation, Transport Friendliness, and Human Readability. Those requirements are very difficult, if not impossible, to meet simultaneously. For example, having a human-readable printed representation makes it possible to cut and paste URCs and send them around via e-mail. Unfortunately, differing conventions for handling EOL, tabs, etc. mean that there will not be a consistent representation for the purposes of digital signatures. These apparently conflicting requirements were addressed in section 5 on multiple transfer syntaxes. It suggests that a variety of transfer syntaxes be allowed. For human readability, a simple text version might be requested. For digital signatures, a binary transfer syntax might be used, and the client would take responsibility for displaying the URC to the user. draft-ietf-uri-urc-sgml-00 [Page 16] INTERNET-DRAFT An SGML-based URC Service June 16, 1995 Simplicity: Utilizing the defaults for the attribute set, and fixing the transfer sytax to something like text/urc0, text/plain, or text/html, URCs can be simple enough for hand entry. Rearrangeability: The rearrangability requirement originated with the need to sort locations according to preferences such as cost, format, size, etc. While this is a very reasonable thing to do, sorting by author name is not so reasonable. The default attribute set is independent of element order. Developers of other attribute sets are encouraged to maintain this flexibility. Developers of software for manipulating URCs are also encouraged to think carefully about what they are doing and not rearrange fields without good reason. Generality Structure Ignorability: The requirements that the URC be general enough to store ANY sort of data, and that it have a self-describing structure, are met by the use of SGML DTDs for URC definition. Being able to determine the extent of novel elements is a function of particular transfer encodings. Searchable: The trivial query language provides enough capability for URN resolution, but it does not provide the general query capability imposed by the searchability requirement. A more capable query language specification is under development. This is discussed further in section 9.1. Subsettable Incrementally Modifiable: It is possible to create new URCs from pieces of other ones. Because all the elements of the default attribute set are optional, any URC prepared according to it can be decimated while still leaving a legal URC. For URCs prepared according to other attribute sets, the attribute set definition must take this requirement into account. There is nothing in this specification that prevents modifying existing elements in a URC or adding new elements. The specification that the semantics of the default attribute set are preserved in all elements named the same as those in teh default attribute set does, however, impose the requirement that novel semantics require a different name. Separable: This is an issue for binary transfer syntaxes. Versioning: Versioning URC changes is still an open issue (see section 9.2). The default attribute set will be extended in the next version of this draft to incorporate versioning information. draft-ietf-uri-urc-sgml-00 [Page 17] INTERNET-DRAFT An SGML-based URC Service June 16, 1995 Caching: Because of the use of HTTP as the resolution protocol, existing HTTP caches and proxy servers can be utilized for caching URC information. Grandfathering: The use of multiple transfer syntaxes and multiple attribute set definitions may ease the grandfathering of older metadata schemes, but this has not been a major consideration in the development of this specification. Some steps in that direction (see section 8.1) have been taken with respect to IAFA templates [6] and the format suggested in RFC 1357 [5] for emailing bibliographic records. 7.2 Requirements on the URC Service Resolution: As shown in section 2, it is possible for the URC service to resolve URNs into URCs. Resolving the URC into a URL can also be performed manually or automatically. The query language also provides the capability for taking a URL and obtaining URCs that contain it, which should allow URL to URN reverse lookups. Multiple Syntaxes: Multiple transfer syntaxes are a feature of this specification. Query Language: As mentioned earlier, the trivial QL in section 6 does not meet all the requirements in [1]. This is discussed further in section 9.1. Security: Because we have used HTTP as the resolution protocol, we believe we can utilize secure HTTP, SSL, and related WWW security work to meet the security requirements. It may be that only particular transfer syntaxes can be secured, but that is deemed an acceptable compromise by the authors of this spec. Authentication Chain: Until we have a reasonable public key infrastructure, this requirement can not truly be met. The system has been architected for the assumption that such a service will arise, and that domain names can be registered as distinguished names in that hierarchy. Do we need to account for the time of a domain name? Access Control: Access control is handled through existing HTTP mechanisms. Future HTTP mechanisms will also be gracefully handled. Maintenance: It is believed that nothing in the URC specification prevents maintenance of the URC information. Particular details of maintenance procedures are for implementations to establish and are outside the scope of this spec. draft-ietf-uri-urc-sgml-00 [Page 18] INTERNET-DRAFT An SGML-based URC Service June 16, 1995 Synchronization: Synchronizing the content of multiple URC servers is outside the scope of this specification. Development: This specification allows multiple URNs to be associated with a single URC, thus meeting the development requirement. Choice: This requirements is more properly addressed as part of the URN resolution specification, but nothing in this specification assumes that users must always go through a particular gateway server. (Note that efficiency considerations may make it very very common to do so, in order to gain the benefits of a caching HTTP proxy, but there is certainly no requirement that we do so). Scalability: This proposal does not forward resolution queries. Am I answering the problem here, or is this more properly a part of various URN resolution specs? Administrative Contact Hierarchical Operations: The trivial query language provides a query for determining administrative information and the structure of the local publication hierarchy. It also provides queries to determine the attribute sets used, which can be used to speed subject-specific resource discovery. 8 Random Notes This section contains notes about the service that are believed useful, but haven't found a graceful home elswhere in the draft. 8.1 Other IETF Metadata Efforts The URC is not the only IETF-sanctioned metadata effort. The IAFA template [6] and the biblio std [5] are other means for conveying information similar to the URC. Because the URC allows different attribute sets to be used, it is possible to incorporate those standards into the URC service. Both IAFA and RFC 1357 define a plain text encoding for their information. Our intent at this time is to require URCs in the text/plain syntax that have an AID of or to conform to those documents. draft-ietf-uri-urc-sgml-00 [Page 19] INTERNET-DRAFT An SGML-based URC Service June 16, 1995 8.2 Referring to Particular URCs Many different URCs can exist for a resource. For example, if a Dept. of Energy laboratory publishes a technical report, they will provide the default URC. Other organizations, such as the Library of Congress, the OCLC, and the DOE's Office of Scientific and Technical Information might all have URCs describing that document. Subject-specific search services would have their own URcs. There will be occasions when we want to indicate that a particular URC should be fetched. This is a job for a URL. As an example, these three URLs show how we might refer to different URCs for the same URN: http://uri.acl.lanl.gov/urn:x-dns-2:uri.acl.lanl.gov/foo57 http://urn.oclc.org/urn:x-dns-2:uri.acl.lanl.gov/foo57 http://urn.gatech.edu/urn:x-dns-2:uri.acl.lanl.gov/foo57 9 Open Issues 9.1 Query Language Requirements say that ``It must be possible to select a URC based on a search of its components. It must be possible to select which components will be searched and which will not be searched.'' They also say ``It must be possible for simple resolution queries to be augmented with information on the version of a resource desired, and an indication of whether signature information should be supplied. It must be possible to select sets of URCs based on criteria such as data of last modification, etc.'' The trivial QL in section 6 does not meet those requirements. Do we need to put a more capable QL into this paper, can we refer to an external draft under development, or do we relax the requirements? I favor the second option. 9.2 Meta-metadata At its simplest, the URC for a resource only contains data about the resource. Of course, this quickly breaks down. The URC can be modified over time, in fact, one of the requirements on the URC service is that we be able to do so. This implies that we might want to know information about the URC - when it was created, by whom, what is its version, what changes were made since the last version and why, etc. Where to put this meta-metadata and how to access it are open issues. Because of the syntax-independence property we wish to provide, URN resolvers will already be constructed with the assumption draft-ietf-uri-urc-sgml-00 [Page 20] INTERNET-DRAFT An SGML-based URC Service June 16, 1995 of dynamically generating the required output. Therefore, it probably makes sense for the resolver to maintain the metadata, meta-metadata, ... locally and only provide it when requested. The default attribute set will probably be augmented with some basic meta-metadata capabilities as this specification is developed further. 9.3 Attribute set object encapsulation Since attribute sets are fundamental to our view of the URC service, it is appropriate to ask what sorts of operations will be defined over them. This is one of our current areas of work. At this time, all the operations we are exposing to the client are "get" methods. Get the complete attribute set, get just the changes from the parent, or the base, get the name of the parent, get the name of the base, get the names of the elements that are new or overridden in the changes, etc. Since the AID is a URN, that means it has a URC, and that URC will also have an attribute set. While we do not want to mandate all the attribute sets - the meta-attribute set may be an appropriate element for standardization. 10 Acknowledgments This specification is the result of many people submitting their ideas to the crucible of forums such as the URI-WG, from whence we have shamelessly plucked them. In particular, we would like to acknowledge Eric Miller of OCLC for providing the original DTD for the Dublin metadata set that has been turned into something he might prefer not to think about on a queasy stomach. Dan Connolly, of the W3O and the HTTP-WG, deserves thanks for his unwitting contribution of the SGML declaration. Many other WG members have contributed to the development of this specification. Particular mention must go to the working group chair, Larry Masinter of Xerox PARC, for the notion of named attribute sets that is central to this specification. Finally, Ron Daniel would like to thank Dave Forslund of Los Alamos National Laboratory and Carol Hunter of Lawrence Livermore National Laboratory for digging out the funding to let him pursue this work. draft-ietf-uri-urc-sgml-00 [Page 21] INTERNET-DRAFT An SGML-based URC Service June 16, 1995 11 References References [1] Ron Daniel and Michael Mealling, ``URC Scenarios and Requirements'', [2] T. Berners-Lee, R.T. Fielding, and H. Frystyk Nielsen, ``Hypertext Transfer Protocol -- HTTP/1.0'', [3] Keith Shafer, Eric Miller, Vincent Tkac, and Stuart Weibel, ``URN Services'', [4] Paul Hoffman and Ron Daniel, ``x-dns-2 URN Scheme'', [5] D. Cohen, ``A Format for E-Mailing Bibliographic Records'', RFC 1357, [6] P. Deutsch, A. Emtage, M. Koster, M. Stumpf, ``Publishing Information on the Internet with Anonymous FTP'', [7] Paul Hoffman and Ron Daniel, ``Trivial URC Syntax: urc0'', [8] Stuart Weibel, Jean Godby, Eric Miller, and Ron Daniel (eds), ``OCLC/NCSA Metadata Workshop Report'', 12 Security Considerations The URC service will provide information of considerable value, especially once Internet payment systems become common. Future versions of this specification must address how transactions can be validated if desired without imposing significant delays on unsecured draft-ietf-uri-urc-sgml-00 [Page 22] INTERNET-DRAFT An SGML-based URC Service June 16, 1995 operations. The author believes that supporting multiple syntaxes will help in this area, and that new attribute sets can be defined which will carry digital signature information. This is addressed in the current version of [1]. Contact Information Ron Daniel Jr. MS B287 Los Alamos National Laboratory Los Alamos, NM, USA 87545 voice: (505) 665-0139 fax: (505) 665-4939 rdaniel@lanl.gov http://www.acl.lanl.gov/ rdaniel/ Terry Allen O'Reilly and Associates, Inc. 101 Morris St. Sebastopol, Calif., 95472 terry@ora.com A SGML Declaration This appendix presents the boring SGML declaration that is needed for specificity. Serious consideration is being given to the use of ISO 10646, we are waiting on the HTML working group to decide what will happen in that area. For now we use the so-called "Latin-1" character set encoding. This SGML declaration was stolen from Dan Connolly's work for the HTML Check package. Hey, I steal from the best. B DTD for Default Attribute Set This appendix presents the default SGML DTD. This is what would be returned as a result of resolving the default AID. (Just what that default AID will be depends on the result of the URN effort, we will assume for now). draft-ietf-uri-urc-sgml-00 [Page 25] INTERNET-DRAFT An SGML-based URC Service June 16, 1995 draft-ietf-uri-urc-sgml-00 [Page 26] INTERNET-DRAFT An SGML-based URC Service June 16, 1995 draft-ietf-uri-urc-sgml-00 [Page 27] INTERNET-DRAFT An SGML-based URC Service June 16, 1995 draft-ietf-uri-urc-sgml-00 [Page 28] INTERNET-DRAFT An SGML-based URC Service June 16, 1995 draft-ietf-uri-urc-sgml-00 [Page 29] INTERNET-DRAFT An SGML-based URC Service June 16, 1995 C DTD for Meta Attribute Set The attribute set definition given in Appendix B is believed to be widely applicable, but is certainly not universally applicable. We encourage others to create their own attribute sets that more closely meet their needs. As described in section 3, since the AID is a URN, there needs to be a URC that talks about the attribute set definition. That URC has particular needs not addressed by the default attribute set. This appendix presents an attribute set definition that is more useful for describing attribute set definitions. When people create their own attribute sets, they are strongly encouraged to use this attribute set to describe them. The AID of this attribute set is for now. IANA should assign a URN once that has been settled. An example of how to use this to create a new attribute set is given in Appendix D. draft-ietf-uri-urc-sgml-00 [Page 30] INTERNET-DRAFT An SGML-based URC Service June 16, 1995 <-- No new elements are defined, and the content models are not changed, so just suck in the default attribute set DTD. The preceding definition of Relationship.Type will override the one in the default DTD. --> %default-as; D Custom Attribute Set Example The role of the meta attribute set definition given in Appendix C is somewhat confusing. Furthermore, we anticipate that most custom attribute sets will actually be small modifications of the default attribute set - merely adding an element or two. This appendix presents a sample DTD to illustrate how to perform such a minor enhancement. It shows a couple of ways of adding a new element, , to the default attribute set. The first way is the simplest. It adds parameter entity definitions to the DOCTYPE declaration in the URC itself. It does not require a new attribute set to be widely published, but has limitations. draft-ietf-uri-urc-sgml-00 [Page 31] INTERNET-DRAFT An SGML-based URC Service June 16, 1995 ] > urn:dns:pchs.k-12.okc.ok.us:student-papers-1995/geo3 Smith, Fred A Vicious, Seditious, and Tendentious History of George III American Revolution (In)famous crackpots of history http://www.pchs.k-12.okc.ok.us/student-papers/1995/smith/geo3.html
text/html
write: fsmith, admin read: any The disadvantage of this approach is that it does not create a named attribute set. The new AS is anonymous. Furthermore, text/sgml is the only syntax that can use this method. The second way to create a new attribute set is to create and publish a new DTD that extends another DTD. The DTD below adds the element to the default attribute set. <-- Suck in the default attribute set's DTD. The n.Instance parameter entity above will override the one in the default attribute set, thereby adding AccessControl to the URC. --> &default-as; If we assume this attribute set has an AID of , a simple URC prepared using it, and transmitted in the text/sgml syntax might look like: urn:dns:pchs.k-12.okc.ok.us:student-papers-1995/geo3 Smith, Fred A Vicious, Seditious, and Tendentious History of George III American Revolution (In)famous crackpots of history http://www.pchs.k-12.okc.ok.us/student-papers/1995/smith/geo3.html
text/html
read:any write:fsmith, admin If we were to resolve the URN of the new attribute set and get back a URC, that URC might look like: draft-ietf-uri-urc-sgml-00 [Page 33] INTERNET-DRAFT An SGML-based URC Service June 16, 1995 urn:x-dns-2:uri.acl.lanl.gov:meta-dtd Daniel, Ron This attribute set adds a stupid access control element to the default attribute set. Don't use this, use someting better. urn:x-dns-2:uri.acl.lanl.gov:default-1-dtd http://www.acl.lanl.gov/attribute-sets/access-ctl-dtd.sgml
text/sgml
 This Internet Draft expires December 15, 1995. draft-ietf-uri-urc-sgml-00 [Page 34]