Copyright © 2002-2003 W3C ® (MIT, ERCIM, Keio), All Rights Reserved. W3C liability, trademark, document use and software licensing rules apply. Your interactions with this site are in accordance with our public and Member privacy statements.
The World Wide Web is a networked information system. Web Architecture consists of the requirements, constraints, principles, and choices that influence the design of the system and the behavior of agents within the system. When Web Architecture is followed, the large-scale effect is that of an efficient, scalable, shared information space. The organization of this document reflects the three divisions of Web architecture: identification, representation, and interaction. This document also addresses some non-technical (social) issues that play a role in building the shared information space.
This document strives to establish a reference set of requirements, constraints, principles, and design choices for Web architecture.
This section describes the status of this document at the time of its publication. Other documents may supersede this document. The latest status of this document series is maintained at the W3C.
This document has been developed by W3C's Technical Architecture Group (TAG) (charter).
This draft is an Editor's Draft and some new content has not yet been reviewed by the TAG. The primary changes in this draft were editorial changes based on some comments from Stuart Williams, and the addition of a note regarding the phrase "on the Web" based on comments from David Orchard. A complete list of changes since the previous Working Draft is available on the Web.
This draft remains incomplete; sections 1, 2, and 3 are the most developed; 4 the least. The TAG has published a number of findings that address specific architecture issues. Parts of those findings may appear in subsequent drafts. Please also consult the list of issues under consideration by the TAG.
This draft includes some editorial notes and also references to open TAG issues. These do not represent all open issues in the document. They are expected to disappear from future drafts.
Publication as a Working Draft does not imply endorsement by the W3C Membership. This is a draft document and may be updated, replaced or obsoleted by other documents at any time. It is inappropriate to cite this document as other than "work in progress."
The latest information regarding patent disclosures related to this document is available on the Web. As of this publication, there are no disclosures.
Please send comments on this document to the public W3C TAG mailing list www-tag@w3.org (archive).
A list of current W3C Recommendations and other technical documents can be found at the W3C Web site.
Highlighted entries in this table of contents link to principles, constraints, good practice notes, and design choices emphasized in the document.
The World Wide Web (or, Web) is a networked information system consisting of agents (programs acting on behalf of a person, entity, or process) that exchange information. Here's a simple travel scenario illustrating a common Web interaction:
http://weather.example.com/oaxaca
" in a
glossy travel magazine. Dan has enough experience with the Web to
recognize that http://weather.example.com/oaxaca
is a
URI. He can expect that the URI should allow him to access relevant
weather information.weather.example.com
.This scenario illustrate the three architectural divisions of the Web that are discussed in this document:
http://weather.example.com/oaxaca
.Editor's note: Todo: Introduce notions of client and server. Relation of client to agent and user agent. Relation of server to resource owner.
The intended audience for this document includes:
This document is designed to balance the value of brevity and precision with the value of illustrative examples. TAG findings provide more background, motivation, and examples.
Readers will benefit from familiarity with the Requests for Comments (RFC) series from the IETF, some of which define pieces of the architecture discussed in this document.
The architecture described in this document is principally the result of experience. There has been some theoretical and modeling work in the area of Web Architecture, notably Roy Fielding's work on "Representational State Transfer" [REST].
The terms MUST, MUST NOT, SHOULD, SHOULD NOT, and MAY are used in accordance with RFC 2119 [RFC2119].
Throughout this document, we elaborate on the travel scenario to introduce and illustrate architectural principles.
Editor's note: The scenario has not yet been well-integrated into sections 3 and 4.
This document focuses on the architecture of the Web. We assume the reader is familiar with the rationale for some of the general design principles: minimal constraints (fewer rules makes the system more flexible), modularity, minimum redundancy, extensibility, simplicity, and robustness.
Other groups inside and outside W3C are writing down principles related to specialized aspects of Web architecture, including accessibility, internationalization, device independence, and Web Services. The section on Architectural Specifications includes some references.
The TAG intends for this document to inform discussions about issues of Web Architecture. Where current practice conflicts with this document, the TAG expects to engage in constructive discussion with other parties. Some parts of this document may fill in gaps in published specifications or may call attention to known weaknesses in those specifications.
This document promotes reuse of existing standards when suitable, and gives some guidance on how to innovate in a manner consistent with the Web architecture.
Uniform Resource Identifiers (URI), defined by "Uniform Resource Identifiers (URI): Generic Syntax" [URI], are central to Web Architecture. Parties who wish to communicate about something will establish a shared vocabulary, i.e. a shared set of bindings between identifiers and things. This shared vocabulary has a tangible value: it reduces the cost of communication. The ability to use common identifiers across communities is what motivates global naming in Web Architecture.
URIs identify resources. When a representation of one resource refers to another resource with a URI, a link is formed between the two resources. The networked information system is built of linked resources, and the large-scale effect is a shared information space. The value of the Web grows exponentially as a function of the number of linked resources (the "network effect").
Principle
Use URIs: All important resources SHOULD be identified by a URI.3
Although there's no precise definition of an "important resource," a resource should have a URI if a third party might reasonably want to link to it, make or refute assertions about it, retrieve or cache a representation of it, transclude all or part of it into another resource, or annotate it.
There are many benefits to making resources identifiable by URI. Some are by design (e.g., linking, bookmarking, and caching), others (e.g., global search services) were not predicted.4
An important aspect of communication is to be able to establish when two parties are talking about the same thing. In the context of the Web, this means when two parties identify the same resource. The most common way to establish that two parties are identifying the same resource is to compare the spelling (i.e., as strings) of the identifiers the parties are using. Section 6 of [URI] discusses this type of analysis. In that specification, determination of equivalence or difference of URIs is based on string comparison, perhaps augmented by reference to additional rules provided by URI scheme definitions (e.g., for HTTP URIs, the authority component is case-insensitive). Depending on the application, an agent may invest more processing effort to reduce the likelihood of a false negative (i.e., two URIs identify the same resource, but that was not detected).
There may be other ways to establish that two parties are identifying the same resource that are not based on string comparison; see the section on future directions for determining that two URIs identify the same resource.
Editor's note: Dan Connolly has suggested the term "coreference" instead of "equivalence" to communicate that two URIs are referring to the same resource.
Agents that reach conclusions about identity beyond what they
are licensed to do (e.g., by specification, or community
convention, or site-specific convention) take responsibility for
any problems that result. For instance, agents should not assume
that http://weather.example.com/Oaxaca
and
http://weather.example.com/oaxaca
identify the same
resource, since none of the specifications involved states that the
path part of an HTTP URI is case-insensitive. Web servers may vary
in how they are configured to handle case-sensitivity. Agents that
assume these URIs identify the same resource take responsibility
for any resulting problems.
Although it is possible to determine that two URIs are equivalent, it is generally not possible by mere inspection of two URIs to be sure that they identify different resources. Web architecture does not constrain resources to be uniquely named.
Good practice
Spelling URIs: If an agent has been provided with a URI to refer to a resource, the agent SHOULD use the spelling of the URI as it was originally provided.
To help parties know when they are referring to the same
resource, it follows that URI producers should be conservative
about the number of different URIs they produce for the same
resource. For instance, the parties responsible for
weather.example.com have no reason to use both
http://weather.example.com/Oaxaca
and
http://weather.example.com/oaxaca
to refer to the same
resource; agents will not detect the equivalence relationship. In
this case, one URI should be chosen and used consistently. See
section 6.3 of [URI] for further
advice on how to reduce the risk of false negatives.
URI consumers cannot, in general, determine the meaning of a
resource by inspection of a URI that identifies it. In our travel scenario, the example URI
(http://weather.example.com/oaxaca
) suggests that the
identified resource has something to do with the weather in Oaxaca.
Although short, meaningful URIs benefit people, URI consumers must
not rely on the URI string to communicate the meaning of a
resource. A site reporting the weather in Oaxaca could just as
easily be identified by the URI
http://vjc.example.com/315
. And the URI
http://weather.example.com/vancouver
might identify
the resource "my photo album." See the section on retrieving a representation for
information about how agents convey information about a
resource.
Editor's note: When finding available on URI opacity, link from here.
In the URI http://weather.example.com/
, the "http"
that appears before the colon (":") is a URI scheme name. There are
other scheme names, such as "mailto" and "ftp". It is common to
classify URIs by scheme; a URI with scheme "http" is called an
"HTTP URI."
Each URI begins with a URI scheme name. The scheme name corresponds to a specification for assigning identifiers within that scheme. As such, the URI syntax is a federated and extensible naming system wherein each scheme's specification may further restrict the syntax and semantics of identifiers using that scheme. Furthermore, the URI scheme specification specifies how an agent can dereference the URI.
Several URI schemes incorporate identification mechanisms that pre-date the Web into this syntax:
mailto:nobody@example.org
ftp://example.org/aDirectory/aFile
news:comp.infosystems.www
tel:+1-816-555-1212
Other URI schemes have been introduced since the advent of the Web, including those introduced as a consequence of new protocols. Examples of URIs for these schemes include:
http://www.example.org/something?with=arg1;and=arg2
ldap://ldap.itd.umich.edu/c=GB?objectClass?one
urn:oasis:SAML:1.0
The Internet Assigned Numbers Authority (IANA) maintains a registry [IANASchemes] of mapping between URI scheme names and their specifications. For instance, the IANA registry indicates that the "http" scheme is defined by [RFC2616]. The process for registration of new URI schemes is defined by RFC2717.
Since many aspects of URI processing are scheme-dependent, and since a huge amount of deployed software already processes URIs of well-known schemes, the cost of introduction of new URI schemes is high. We note in passing that even more expensive than introducing a new URI scheme is introducing a new identification mechanism for the Web; this is considered prohibitively expensive.
Good practice
New URI schemes: Authors of specifications SHOULD avoid introducing new URI schemes when existing schemes can be used to meet the goals of the specifications.
Consider our travel
scenario: should the authority providing information about the
weather in Oaxaca register a new URI scheme "weather" for the
identification of resources related to the weather? They might then
publish URIs such as
weather://travel.example.com/oaxaca
. While the Web
Architecture allows the definition of new schemes, there is a cost
to registration and especially deployment of new schemes. When an
agent dereferences such a URI, if what really happens is that HTTP
GET is invoked to retrieve an HTML representation of the resource,
then an HTTP URI would have sufficed. If a URI scheme exists that
meets the needs of an application, designers should use it rather
than invent one. Furthermore, designers should expect that it will
prove useful to be able to share a URI across applications, even if
that utility is not initially evident.
If the motivation behind registering a new scheme is to allow an agent to launch a particular application when retrieving a representation, such dispatching can be accomplished at lower expense by registering a new media type instead. Reasons for this include:
Editor's note: When finding available based on Tim Bray's discussion of this topic, link from here.
The use of unregistered URI schemes is discouraged for a number of reasons:
In the URI http://weather.example.com/
, the string
weather.example.com
(between "//" and the next "/")
called the authority component. Many URI schemes include a
hierarchical element for a naming authority such that governance of
the name space defined by the remainder of the URI is delegated to
that authority (which may, in turn, delegate it further). The
generic syntax provides a common means for distinguishing an
authority based on a registered domain name or server address. See
section 3.2 of [URI] for more
information about the authority portion of a URI.
How authority is delegated depends on the URI scheme. The deployment and use of different URI schemes may require varying degrees of central coordination and administration. For example, MAILTO, FTP, and HTTP URIs depend on the use of the DNS and IANA infrastructure; see "ICP-1: Internet Domain Name System Structure and Delegation" [IANAICP1] for more information about how the IANA manages delegation of domain names.
Successful communication between two parties about a piece of information relies on shared understanding of the meaning of the information. Thousands of independent parties can identify and communicate about a Web resource. To give these parties the confidence that they are all talking about the same thing when they refer to "the resource identified by the following URI ..." the design choice for the Web is, in general, that the owner of a resource assigns the authoritative interpretation of representations of the resource. See the draft TAG finding "Client handling of MIME headers" for related discussion.
In our travel scenario, the
agent responsible for weather.example.com
has license
to create representations of this resource and assign their
authoritative interpretation.
In our travel scenario the
server returns an XHTML representation when Dan dereferences the
URI http://weather.example.com/oaxaca
. Then, by
navigating within the XHTML content, Dan finds that the URI
http://weather.example.com/oaxaca#tom
refers to
information about tomorrow's weather in Oaxaca. This URI includes
the fragment identifier "tom" (the string after the "#").
The fragment identifier component of a URI allows indirect identification of a secondary resource, by reference to a primary resource and additional identifying information that is selective with respect to that resource. The identified secondary resource may be some portion or subset of the primary resource, some view on representations of the primary resource, or some other resource that is merely named with respect to the primary resource.
Although the generic URI syntax allows any URI to end with a fragment identifier, some URI schemes do not specify the use of fragment identifiers. For instance, fragment identifier usage is not specified for MAILTO URIs.
For URI schemes that do specify the use of fragment identifiers, the syntax and semantics of those identifiers is defined by the set of representations that might result from a retrieval action on the primary resource. The presence of a fragment identifier component in a URI does not imply that a retrieval action will take place.
Interpretation of the fragment identifier during a retrieval action is performed solely by the user agent; the fragment identifier is not passed to other systems during the process of retrieval. This means that some intermediaries in the Web architecture (e.g., proxies) have no effect on fragment identifiers and that redirection (in HTTP [RFC2616], for example) does not account for them.
Suppose that the managers of weather.example.com
provide a visual map of the meteorological conditions in Oaxaca as
part of the representation served for
http://weather.example.com/oaxaca
. They might encode
the same visual map in a number of image formats to meet different
needs (e.g., they might serve PNG, SVG, and JPEG/JFIF). Dan's user
agent and the server engage in HTTP content negotiation, so that
Dan receives the best image format his user agent can handle or the
format he usually prefers. The URI
http://weather.example.com/oaxaca/map#zicatela
refers
to a portion of the weather map that shows the Zicatela Beach,
where Dan intends to go surfing. Clients can do something useful
with the fragment identifier and the SVG representation, since SVG
defines fragment identifier semantics. Clients should not be
expected to do something useful with the fragment identifier for
the PNG or JPEG/JFIF representations since those specifications do
not define fragment identifier semantics.
Good practice
Content negotiation with fragments: Authors SHOULD NOT use HTTP content negotiation for different media types that have incompatible fragment identifier semantics.
Given a URI, a system may attempt to perform a variety of operations on the resource, as might be characterized by such words as "access", "update", "replace", or "find attributes". Available operations depend on the formats and protocols that make use of URIs. The URI specification (in [URI], section 1.2.2) defines the following terms related to interactions through a URI.
During URI resolution, an agent applies in succession a finite set of relevant specifications, beginning with the specification of the context in which the URI is found (e.g., a format or protocol specification, or an application). Any one of these specifications may define more than one access mechanism (e.g., the HTTP protocol defines a number of access methods, including GET, HEAD, and POST). Note that the information governing the choice of access mechanism may be found in the context, not the URI itself (e.g., the choice of HTTP GET v. HTTP HEAD). The draft TAG finding "URIs, Addressability, and the use of HTTP GET and POST." discusses issues surrounding multiple access mechanisms and the relation to URI addressability.
Some URI schemes (e.g., the URN scheme [RFC 2141]) do not define dereference mechanisms.
TAG issue metadataInURI-31: Should metadata (e.g., versioning information) be encoded in URIs?
TAG issue siteDate-26: Web site metadata improving on robots.txt, w3c/p3p and favicon etc.
One of the most important actions involving a resource is to retrieve a representation of it (for example, by using HTTP GET). As stated above, the authority responsible for a URI determines what the URI identifies and which representations are used for interaction with the resource. The representations communicate the state of the resource.
Good practice
Resource descriptions: Owners of important resources SHOULD make available representations that describe those resources.
As an example of representation retrieval, suppose that the URI
http://weather.example.com/budapest
is used within an
a
element of an SVG document. The sequence of
specifications applied is:
a
link involves retrieving a representation of a
resource, identified by the XLink href
attribute: "By
activating these links (by clicking with the mouse, through
keyboard input, and voice commands), users may visit these
resources."xlink:href
is defined in section 5.4
of the XLink 1.0 [XLink10]
specification states that "The value of the href attribute must be
a URI reference as defined in [IETF RFC 2396], or must result in a
URI reference after the escaping procedure described below is
applied."Note that, in general, one cannot determine the media type(s) of
representation(s) of a resource by inspecting a URI for that
resource. For example, do not assume that all representations of
http://example.com/page.html
are HTML. The HTTP
protocol does not constrain the media type based on the path
component of the URI; the server is free to return a PNG image
representation.
Editor's note: It is an open question whether the TAG will define and use the phrase "on the Web" in this document. Definitions that have been suggested include "is identified by a URI" and "is identified by a URI and at least one representation is available for retrieval."
Dan's retrieval of weather information qualifies as a "safe" interaction; a safe interaction is one where the user agent does not commit to anything beyond the interaction and is not responsible for any consequences other than the interaction itself (e.g., a read-only query or lookup). Other Web interactions resemble orders more than queries. These unsafe interactions may cause a change to the state of a resource; the user may be held responsible for the consequences of these interactions. Unsafe interactions include subscription services, posting to a list, or modifying a database.
Safe interactions are important because these are interactions where users can browse with confidence and where software programs (e.g., search engines and browsers that pre-cache data for the user) can follow links safely. Users (or software agents acting on their behalf) do not commit themselves to anything by querying a resource or following a link.
Principle
Safe retrieval: Agents do not incur obligations by retrieving a representation.
For instance, suppose in our travel scenario that the managers of
weather.example.com
offer a monthly newsletter
available by subscription. It is incorrect and
harmful to publish a page
http://example.com/oxaca/aboutNewsLetter
that states
"... terms and conditions..." with a link to
http://example.com/oxaca/newsLetter
because search
services may link directly to
http://example.com/oxaca/newsLetter
and readers that
follow such links may not have seen, let alone agreed to, the terms
and conditions.
For more information about safe and unsafe operations using HTTP GET and POST, and handling security concerns around the use of HTTP GET, see the draft TAG finding "URIs, Addressability, and the use of HTTP GET and POST."
The value of a URI increases with the predictability of interactions using that URI.
Good practice
URI Persistence: Parties responsible for a URI SHOULD service that URI predictably and consistently.
Service breakdowns include:
There are strong social expectations that once a URI identifies a particular resource, it should continue indefinitely to refer to that resource; this is called URI persistence. URI persistence is always a matter of policy and commitment on the part of authorities servicing URIs rather than a constraint imposed by technological means.
URI persistence also improves when ambiguity is removed about
what the URI identifies. For instance, saying that the URI
http://www.example.com/moby
identifies "Moby Dick" can
lead to confusion because this might be interpreted as any one of
the following very distinct resources: a particular printing of
this work (say, by ISBN), or the work itself in an abstract sense,
or the fictional white whale, or a particular copy of the book on
the shelves of a library (via the Web interface of the library's
online catalog), or the record in the library's electronic catalog
which contains the metadata about the work, or the Gutenberg
project's online
version. Similarly, one should not use the same URI to refer to
a person and to that person's mailbox.
Ambiguous descriptions of what a URI identifies increase the likelihood that two parties will think the same URI identifies different resources, and thus that the parties will use the URI inconsistently. This can be costly, as in the case of two databases in which the same URI is used inconsistently; merging the two databases might lead to confusion or errors.
HTTP [RFC2616] has been designed to help service URIs. For example, HTTP redirection (via some of the 3xx response codes) permits servers to tell an agent that further action needs to be taken by the agent in order to fulfill the request (e.g., the resource has been assigned a new URI). In addition, content negotiation also promotes consistency, as a site manager would not be required to define new URIs for each new format that is supported, as would be the case with protocols that don't support content negotiation, such as FTP.
For more discussion about URI persistence, refer to [Cool].5
As we have seen, identification of a resource is distinct from interacting with that resource. It is reasonable to control access to the resource (e.g., for security reasons), but it is unreasonable to prohibit others from merely identifying the resource.
As an analogy: A building might have a policy that the public may only enter via the main front door, and only during business hours. People employed in the building and in making deliveries to it might use other doors as appropriate. Such a policy would be enforced by a combination of security personnel and mechanical devices such as locks and pass-cards. One would not enforce this policy by hiding some of the building entrances, nor by requesting legislation requiring the use of the front door and forbidding anyone to reveal the fact that there are other doors to the building.
In the travel scenario,
imagine that Dan and Norm both subscribe to the
weather.example.com
newsletter. Dan wishes to point
out an article of particular interest to Norm, using a URI. The
managers of weather.example.com
can offer Dan and Norm
the benefits of URIs (e.g., bookmarking and linking) and still
control access to the newsletter by authorized parties. The Web
provides several mechanisms to control access to resources, none of
which relies on hiding or suppressing URIs for those resources. For
more information on identification and access control, please refer
to the TAG finding "'Deep
Linking' in the World Wide Web."
The integration of internationalized identifiers (i.e., composed of characters beyond those allowed by [URI]) into the Web Architecture is an important and open issue. See TAG issue IRIEverywhere-27 for discussion about work going on in this area.
Emerging Semantic Web technologies, including "DAML+OIL" [DAMLOIL] and "Web Ontology
Language (OWL)" [OWL10], define
RDF properties such as equivalentTo
and
FunctionalProperty
to state -- or at least claim --
formally that two URIs identify the same resource.
There has been some discussion but no agreement that new access protocols should provide a means to convert fragment identifiers according to media type.
Fragment identifier semantics may differ among formats. See related TAG issues httpRange-14 and RDFinXHTML-35 and abstractComponentRefs-37.
The Dynamic Delegation Discovery System (DDDS) ([RFC3401] and related RFCs) is used to implement lazy binding of strings to data, in order to support dynamically configured delegation systems. This system is designed to allow resolution of any type of URI, in particular URNs.
One area of work involves the creation of globally unique identifiers in a file-sharing system without centralized or hierarchical administration.
A representation is data that represents or describes the state of a resource. It consists of:
Web agents use representations to modify as well as retrieve resource state.
When agents transfer representations via messages (see the section on interactions for information about message protocols), the message often includes additional metadata that is not part of the representation (e.g., the HTTP Server field or the request method, URI, etc.). A message may even include "meta-metadata" (for message-integrity checks). The representation consists those bits that would not change regardless of the transfer protocol used to exchange them.
In our previous travel scenario the representation Dan receives (and whether he receives one at all) depends on a number of factors, including:
weather.example.com
respond to requests at all;weather.example.com
make available one or more
representations for the resource identified by
http://weather.example.com/oaxaca;
weather.example.com
have provided more than one representation (in different formats
such as HTML, PNG, or RDF, in different languages such as English
and Spanish, etc.), the resulting representation may depend on
negotiation between the user agent and server that occurs as part
of the HTTP transaction.We discuss these issues in more detail below.
As discussed above, the owner of a resource assign URIs for that resource, create representations of the resource, and assign their authoritative interpretation. This interpretation is described in part by metadata that is part of the representation, notably the Internet media type. At times there may be inconsistencies between metadata and what is specified in a format. Examples of inconsistencies between headers and format data that have been observed in practice include:
User agents should detect such inconsistencies but should not resolve them without involving the user (e.g., by securing permission or at least providing notification). User agents must not silently ignore authoritative server metadata.
See the draft TAG finding "Client handling of MIME headers" for more in-depth discussion and examples.
The Web can be used to interchange resource representations in any format. This flexibility is important, since there is continuing progress in the development of new data formats for new applications and the refinement of existing ones.
For a format to be usefully interoperable between two parties, the parties must have a shared understanding of its syntax and semantics. This is not to imply that a sender of data can count on constraining its treatment by a receiver; simply that making good use of electronic data usually requires knowledge of its designers' intentions.
For a format to be widely interoperable across the Web:
Although the Web architecture allows for the deployment of new data formats, the creation and deployment of new formats (and software able to handle them) can be very expensive. Thus, before inventing a new data format, designers should carefully consider re-using one that is already available. For example, if a format is required to contain human-readable text with embedded hyperlinks, it is almost certainly better to use HTML for this purpose than to invent a new format.
As noted above, the utility of data formats starts with the availability of a normative specification. Some of the desirable characteristics of a format include:
The section on architectural specifications includes references to additional format specification guidelines.
Other design issues:
This section discusses important characteristics of data formats which can together be used to describe and understand them.
A textual data format is one in which the data is specified as a linear sequence of characters. HTML, Internet e-mail, and all XML-based languages are textual. In modern textual data formats, the characters are usually taken from the Unicode repertoire.
Binary data formats are those in which portions of the data are encoded for direct use by computer processors, for example thirty-two bit little-endian two's-complement and sixty-four bit IEEE double-precision floating-point. The portions of data so represented include numeric values, pointers, and compressed data of all sorts.
In principle, all data can be represented using textual formats.
The trade-offs between binary and textual data formats are complex and application-dependent. Binary formats can be substantially more compact, particularly for complex pointer-rich data structures. Also, they can be consumed more rapidly by software in those cases where they can be loaded into memory and used with little or no conversion.
Textual formats are often more portable and interoperable, since there are fewer choices for representation of the basic units (characters), and those choices are well-understood and widely implemented.
Textual formats also have the considerable advantage that they can be directly read and understood by human beings. This can simplify the tasks of creating and maintaining processing software, and allow the direct intervention of humans in the processing chain without recourse to tools any more complex than the ubiquitous text editor. Finally, it simplifies the necessary human task of learning about new data formats (the "View Source" effect).
All things being equal (a rare state of affairs) textual formats are generally preferable to binary ones in Web applications.
It is important to emphasize that intuition as to such matters as data size and processing speed are not a reliable guide in data format design; quantitative studies are essential to a correct understanding of the trade-offs.
TAG issue binaryXML-30: Effect of Mobile on architecture - size, complexity, memory constraints. Binary infosets, storage efficiency.
Final-form data formats are not designed to allow modification or uses other than that intended by their designers. An example would be PDF, which is designed to support the presentation of page images on either screen or paper, and is not readily used in any other way. XSL Formatting Objects (XSL-FO) share this characteristic.
XHTML, on the other hand, can be and is put to a variety of uses including direct display (with highly flexible display semantics), processing by network-sensitive Web spiders to support search and retrieval operations, and reprocessing into a variety of derivative forms.
In general XML-based data formats are more re-usable and repurposable than the alternatives, although the example of XSL-FO shows that this is not an absolute.
There are many cases where final-form is an application requirement; representations which embody legally-binding transactions are an obvious example. In such cases, the use of digital signatures may be appropriate to achieve immutability, whether the format is naturally final-form or some XML vocabulary.
On the other hand, where such requirements are not in play, representations that are reusable and repurposable are in general higher in value, particularly in the case where the information's utility may be long-lived.
Some data formats are explicitly designed to be used in combination with others, while some are designed for standalone use. An example of a standalone data format is PDF; it is not typically embedded in representations encoded in other formats.
At the other extreme is SOAP, which is designed explicitly to contain a "payload" in some non-SOAP vocabulary. Another example is SVG, which is designed to be included in compound documents, and which may in turn contain information encoded in other XML vocabularies.
This characteristic is related to, but distinct from, the final-form/reusable distinction discussed above. For example, one can certainly imagine cases where it is useful for a representation to include data in multiple different formats, but be considered immutable and display-only.
TAG issue xmlProfiles-29: When, whither and how to profile W3C specifications in the XML Family?
TAG issue mixedUIXMLNamespace-33: Composability for user interface-oriented XML namespaces
TAG issue xmlFunctions-34: XML Transformation and composability (e.g., XSLT, XInclude, Encryption)
TAG issue RDFinXHTML-35: Syntax and semantics for embedding RDF in XHTML
XML and XML Namespaces allow format designers to create and combine vocabularies. A format is extensible if instances of the format can include terms from other vocabularies.
Good practice
Extensibility: Format designers should create extensible formats.
Versioning is the term for the evolution of languages and documents. Versioning is achieved through extensibility mechanisms and language redefinition. When a format definition changes (e.g., by addition or removal of element or attribute definitions), this creates a new version of the format. When an instance of a format includes other vocabulary elements, this creates an extension of the instance; the original format definition has not changed. An example is a SOAP message with a header block, this is called a SOAP extension, not a new version of the SOAP format.
The following terms define important relationships among different versions of a format:
Good practice
Compatibility: Format designers SHOULD define extensibility models that allow forwards compatible and backwards compatible changes.
Naturally, even if M and N are compatible but different versions of a format, agents will process instances of them differently. For instance, if format version M is forwards compatible with format version N, M processors that encounter N instances might handle unknown elements by ignoring them entirely, or by ignoring element tags but continuing to process element content. Different format specifications may require different compatibility behavior.
Good practice
Compatibility behavior: Format designers SHOULD define expected behavior when agents designed to process one version of a format encounter a compatible version of the format.
In some cases, format designers require that new features be
supported (i.e., not ignored). In this case, the new version of the
format (N) may be backwards compatible with the earlier version
(M), but M is not forwards compatible with N. The SOAP 1.2
Recommendation [SOAP12], for
example, defines the mustUnderstand
attribute in section
5.2.3.
For more information on format extensibility, refer to "Web Architecture: Extensible Languages" [EXTLANG].
In many cases, the information contained in a separation is logically separable from the choice of ways in which it may be presented to a human, and the modes of interaction it may support.
While such separation is, where possible, often advantageous, it is clearly not always possible and in some cases not desirable either.
Replacement text from C. Lilley expected.
One of the greatest strengths of HTML as a format is that it allows authors to embed cross references (hyperlinks). The simplicity of <a href="#foo"> as a link to "foo" and <a name="foo"> as the anchor "foo" are partly (perhaps largely) responsible for the birth of the hypertext Web as we know it today.
Simple, single-ended, single-direction, inline links are not the most powerful linking paradigm imaginable. But they are very easy to understand. And they can be authored by individuals (or other agents) that have no control or write access to the other end point.
More sophisticated linking mechanisms have been invented for the Web. XPointer allows links to address content that does not have an explicit, named anchor. XLink allows links to have multiple ends and to be expressed either inline or in "link bases" stored external to any or all of the resources identified by the links it contains.
All of the current common linking mechanisms identify resources by URI and optionally identify portions (or views) of a resource with the fragment identifier.
Good practice
Link mechanisms: Format specification designers SHOULD provide mechanisms for identifying links to other resources and to portions of representations (via fragment identifiers).
For formats based on XML, format designers should examine XLink and the XPointer framework for inspiration. To define fragment identifier syntax, use at least the XPointer Framework and XPointer element() Schemes.
If a future revision of RFC 3023 identifies the XPointer Framework, element(), and perhaps other ancillary schemes as the fragment identifier syntax for XML documents, authors will be able to rely on at least those schemes for all XML documents.
TAG issue: What is the scope of using XLink? xlinkScope-23.
Many resource representations are encoded in formats which are XML vocabularies. This section discusses issues that are specific to such data formats.
Anyone seeking guidance in this area is urged to consult the "Guidelines For The Use of XML in IETF Protocols" [IETFXML] for the use of XML in Internet Protocols. This document contains a very thorough discussion of the considerations that govern whether or not XML ought to be used, as well as specific guidelines on how it ought to be used. While it is directed at Internet applications with specific reference to protocols, the discussion is generally applicable to Web scenarios as well.
The discussion here should be seen as ancillary to the content of the IETF BCP. Refer also to "XML Accessibility Guidelines" [XAG] for help designing XML formats that lower barriers to Web accessibility for people with disabilities.
XML defines textual data formats that are naturally suited to describing data objects which are hierarchical and processed in an in-order sequence. It is widely but not universally applicable for format specifications. For example, an audio or video format is unlikely to be well suited to representation in XML. Design constraints that would suggest the use of XML include:
Editor's note: Which XML Specifications make up the XML Family?
The Web is significantly a networked information system. Authors and applications can use URIs uniformly to identify different resources. After representations of these resources have been retrieved, they may be processed in a variety of ways. Some applications (and some users) will undoubtedly build new resources by combining several representations together. This is particularly easy, and potentially useful, when XML representations are available for all the resources.
However, combining representations in this way moves them out of their original context and places them in a new context. This change of context introduces the possibility of information loss. Any information that depended on the local context will no longer be available.
What is needed is a mechanism for establishing a global context for the elements and attributes in the XML resources. This problem bears a strong resemblance to the distinction between relative and absolute URIs. While the many hundreds of relative URI references to "index.html" on a typical web server may be entirely unambiguous in their respective contexts, they have no unambiguous global meaning. But each such relative URI has an unambiguous absolute URI that can be established in its local context and used when a document is moved. This solves the problem for URI references.
For elements and attributes, their names can be seen as analogous to relative URI. Within their original context, they have meanings that are clear and entirely unambiguous. Namespaces in XML provides a mechanism for establishing a globally unique name that can be understood in any context.
The "absolute" form of an XML element or attribute name is the combination of its namespace URI and its local name. This is represented lexically in documents by associating namespace names with (optional) prefixes and combining prefixes and local names with a colon as described in "Namespaces in XML" [XMLNS].
Designers that use namespaces are thus providing a global context for documents authored with their schema. Establishing this global context allows their documents (and portions of their documents) to be reused and combined in novel ways not yet imagined. Failure to provide a namespace makes such reuse more difficult, perhaps impractical in some cases.
The most significant technical drawback to using namespaces is that they do not interact well with DTDs. DTDs perform validation based on the lexical form of the name, making prefixes semantically significant in ways that are not desirable. As other schema language technologies become widely deployed, this drawback will diminish in significance.
Namespace designers SHOULD make available human-readable material to meet the needs of those who will use the namespace vocabulary. The simplest way to achieve this is for the namespace name to be an HTTP URI which may be dereferenced to access this material. The resource identified by such a URI is called a "namespace document."
There are many reasons why a person or agent might want more information about the namespace. A person might want to:
A namespace document should also support the automatic retrieval of other Web resources that support the processing markup from this vocabulary. Useful information to processors includes:
It follows that there is, in general, no single type of resource that can be returned in response to a request for the namespace name that will always be the most appropriate; see the section on future work regarding namespace document formats for more information.
Issue: namespaceDocument-8: What should a "namespace document" look like?
Issue: abstractComponentRefs-37: Definition of abstract components with namespace names and frag ids
Editor's note: Where should we put a section on mixing namespaces; is the section on processing model more appropriate? See issue mixedUIXMLNamespace-33.
Suppose that the URI http://example.com/oaxaca
defines a resource with representations encoded in XML. What, then,
is the interpretation of the URI
http://example.org/oaxaca#weather
?
The URI specification [URI] makes it clear that the interpretation depends on the context of the media type of the representation. It follows from this that designers of XML-based data formats SHOULD include the semantics of fragment identifiers in their designs. The XPointer Framework [XPTRFR] provides a syntax designed for in such fragment identifiers, and it SHOULD be used for this purpose.
When a representation is provided whose media type is
application/xml
, there are no semantics defined for
fragment identifiers, and thus they SHOULD NOT be provided for such
representations. This is also the case if the representation is
known to be XML because the media type has a suffix of
+xml
as described in "XML Media Types" [RFC3023], but there is no
normative specification of fragment semantics.
It is common practice to assume that when an element has an
attribute that is declared in a DTD to be of type ID, then the
fragment identifier #abc
identifies the element which
has an attribute of that type whose value is "abc"
.
However, there is no normative support for this assumption and it
is problematic in practice, since the only defined way to establish
that an attribute is of type ID is via a DTD, which may not exist
or may not be available.
TAG issue fragmentInXML-28: Do fragment identifiers refer to a syntactic element (at least for XML content), or can they refer to abstractions? See TAG issue.
TAG issue xmlIDSemantics-32: How should the problem of identifying ID semantics in XML languages be addressed in the absence of a DTD? See draft TAG finding "How should the problem of identifying ID semantics in XML languages be addressed in the absence of a DTD?".
RFC 3023 defines the media types application/xml
and text/xml
, and describes a convention whereby
XML-based data formats use media types with a +xml
suffix, for example image/svg+xml
.
In general, media types beginning with text/
SHOULD
NOT be used for XML representations. They create two problems:
First, intermediate agents in the Web are allowed to "transcode",
i.e., convert one character encoding to another. Since XML
documents are designed to allow them to be self-describing, and
since this is a good and widely-followed practice, any such
transcoding will make the self-description false.
Secondly, representations whose media types begin with
text/
are required, unless the charset
parameter is specified, to be considered to be encoded in US-ASCII.
In the case of XML, since it is self-describing, it is good
practice to omit the charset
parameter, and since XML
is very often not encoded in US-ASCII, the use of
"text/
" media types effectively precludes this good
practice.
This section will describe the architectural principles and constraints regarding interactions between components, including such topics as network protocols and interaction styles, along with interactions between the Web as a system and the people that make use of it. This will include the role of architectural styles, such as REST and SOAP, and the impact of meta-architectures, such as Web Services and the Semantic Web.
Good practice
Glossary not yet completed.
Editor's note: The TAG is still experimenting with the categorization of points in this document. This list is likely to change. It has also been suggested that the categories clearly indicate their primary audience.
The important points of this document are categorized as follows:
*
had been chosen
instead, the large-scale result would, most likely, have been the
same. Other design choices are more fundamental; these are the
focus of this document.Editor's note: The usage of a normative reference in this document needs clarification.
The authors of this document are the participants of W3C's Technical Architecture Group: Tim Berners-Lee (Chair, W3C), Tim Bray (Antarctica Systems), Dan Connolly (W3C), Paul Cotton (Microsoft), Roy Fielding (Day Software), Chris Lilley (W3C), David Orchard (BEA Systems), Norman Walsh (Sun), and Stuart Williams (Hewlett-Packard).
The TAG thanks people for their thoughtful contributions on the TAG's public mailing list, www-tag (archive).