Binding Extensions to Web Distributed Authoring and Versioning (WebDAV)IBM20 Maguire RoadLexingtonMA02421geoffrey.clemm@us.ibm.comIBM ResearchP.O. Box 704Yorktown HeightsNY10598ccjason@us.ibm.comgreenbytes GmbHSalzmannstrasse 152MuensterNW48159Germanyjulian.reschke@greenbytes.deUC Santa Cruz, Dept. of Computer Science1156 High StreetSanta CruzCA95064ejw@cse.ucsc.edu
This specification defines bindings, and the BIND method for creating
multiple bindings to the same resource. Creating a new binding to a
resource causes at least one new URI to be mapped to that resource. Servers
are required to insure the integrity of any bindings that they
allow to be created.
Please send comments to the
Distributed Authoring and Versioning (WebDAV) working group at , which may be joined by sending a message with subject
"subscribe" to .
Discussions of the WEBDAV working group are archived at
.
lists all registered issues since draft 02.
Umbrella issue for editorial fixes/enhancements.
This specification extends the WebDAV Distributed Authoring Protocol to enable
clients to create new access paths to existing resources. This capability is
useful for several reasons:
URIs of WebDAV-compliant resources are hierarchical and correspond to a hierarchy
of collections in resource space. The WebDAV Distributed Authoring Protocol
makes it possible to organize these resources into hierarchies, placing them
into groupings, known as collections, which are more easily browsed and
manipulated than a single flat collection. However, hierarchies require
categorization decisions that locate resources at a single location in the
hierarchy, a drawback when a resource has multiple valid categories. For
example, in a hierarchy of vehicle descriptions containing collections for
cars and boats, a description of a combination car/boat vehicle could belong
in either collection. Ideally, the description should be accessible from both.
Allowing clients to create new URIs that access the existing resource lets
them put that resource into multiple collections.
Hierarchies also make resource sharing more difficult, since resources that
have utility across many collections are still forced into a single collection.
For example, the mathematics department at one university might create a
collection of information on fractals that contains bindings to some local
resources, but also provides access to some resources at other universities. For
many reasons, it may be undesirable to make physical copies of the shared
resources on the local server: to conserve disk space, to respect copyright
constraints, or to make any changes in the shared resources visible
automatically. Being able to create new access paths to existing resources in
other collections or even on other servers is useful for this sort of case.
The BIND method defined here provides a mechanism for allowing clients to
create alternative access paths to existing WebDAV resources. HTTP
and WebDAV
methods are able to work because there are mappings between URIs and resources. A
method is addressed to a URI, and the server follows the mapping from that
URI to a resource, applying the method to that resource. Multiple URIs may be
mapped to the same resource, but until now there has been no way for clients
to create additional URIs mapped to existing resources.
BIND lets clients associate a new URI with an existing WebDAV resource, and
this URI can then be used to submit requests to the resource. Since URIs of
WebDAV resources are hierarchical, and correspond to a hierarchy of
collections in resource space, the BIND method also has the effect of adding
the resource to a collection. As new URIs are associated with the resource,
it appears in additional collections.
A BIND request does not create a new resource, but simply makes available a
new URI for submitting requests to an existing resource. The new URI is
indistinguishable from any other URI when submitting a request to a resource. Only
one round trip is needed to submit a request to the intended target.
Servers are required to enforce the integrity of the relationships between
the new URIs and the resources associated with them. Consequently, it may be
very costly for servers to support BIND requests that cross server boundaries.
This specification is organized as follows. defines terminology
used in the rest of the specification, while overviews bindings.
defines the new properties needed to support multiple bindings to
the same resource. specifies the BIND method, used to create
multiple bindings to the same resource. specifies the UNBIND
method, used to remove a binding to a resource. specifies the
REBIND method, used to move a binding to another collection.
The terminology used here follows and extends that in the WebDAV Distributed
Authoring Protocol specification .
The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", "SHOULD",
"SHOULD NOT", "RECOMMENDED", "MAY", and "OPTIONAL" in this document are to be
interpreted as described in .
This document uses XML DTD fragments
() as a purely notational convention.
WebDAV request and response bodies cannot be validated due to the specific
extensibility rules defined in section 23 of and due to the fact
that all XML elements defined by this specification use the XML namespace
name "DAV:". In particular:
Element names use the "DAV:" namespace.Element ordering is irrelevant.Extension elements/attributes (elements/attributes not already defined
as valid child elements) may be added anywhere, except when explicitly
stated otherwise.
URI Mapping
A relation between an absolute URI and a resource. For an absolute URI U
and the resource it identifies R, the URI mapping can be thought of as
(U => R). Since a resource can represent items that are not network
retrievable, as well as those that are, it is possible for a resource to
have zero, one, or many URI mappings. Mapping a resource to an "http"
scheme URI makes it possible to submit HTTP protocol requests to the
resource using the URI.
Path Segment
Informally, the characters found between slashes ("/") in a URI.
Formally, as defined in section 3.3 of .
Binding
A relation between a single path segment (in a collection) and a resource. A
binding is part of the state of a collection. If two different
collections contain a binding between the same path segment and the same
resource, these are two distinct bindings. So for a collection C, a path
segment S, and a resource R, the binding can be thought of as C:(S -> R).
Bindings create URI mappings, and hence allow requests to be sent to a
single resource from multiple locations in a URI namespace. For example,
given a collection C (accessible through the URI
http://www.example.com/CollX), a path segment S (equal to "foo.html"), and
a resource R, then creating the binding C: (S -> R) makes it possible to
use the URI http://www.example.com/CollX/foo.html to access R.
Collection
A resource that contains, as part of its state, a set of bindings that
identify internal member resources.
Internal Member URI
The URI that identifies an internal member of a collection, and that
consists of the URI for the collection, followed by a slash character
('/'), followed by the path segment of the binding for that internal member.
In , the state of a collection is defined as containing a list of
internal member URIs. If there are multiple mappings to a collection, then
the state of the collection is different when you refer to it via a different
URI. This is undesirable, since ideally a collection's membership should
remain the same, independent of which URI was used to reference it.
The notion of binding is introduced to separate the final segment of a URI
from its parent collection's contribution. This done, a collection can be
defined as containing a set of bindings, thus permitting new mappings to a
collection without modifying its membership. The authors of this
specification anticipate and recommend that future revisions of
will update the definition of the state of a collection to correspond to the
definition in this document.
A "precondition" of a method describes the state on the server that must be
true for that method to be performed. A "postcondition" of a method
describes the state on the server that must be true after that method has
completed. If a method precondition or postcondition for a request is not
satisfied, the response status of the request MUST be either 403 (Forbidden)
if the request should not be repeated because it will always fail, or 409
(Conflict) if it is expected that the user might be able to resolve the
conflict and resubmit the request.
In order to allow better client handling of 403 and 409 responses, a
distinct XML element type is associated with each method precondition and
postcondition of a request. When a particular precondition is not satisfied
or a particular postcondition cannot be achieved, the appropriate XML element
MUST be returned as the child of a top-level DAV:error element in the
response body, unless otherwise negotiated by the request. In a 207
Multi-Status response, the DAV:error element would appear in the
appropriate DAV:responsedescription element.
Bindings are part of the state of a collection. They define the internal
members of the collection, and the names of those internal members.
Bindings are added and removed by a variety of existing HTTP methods. A
method that creates a new resource, such as PUT, COPY, and MKCOL, adds a
binding. A method that deletes a resource, such as DELETE, removes a
binding. A method that moves a resource (e.g. MOVE) both adds a binding
(in the destination collection) and removes a binding (in the source
collection). The BIND method introduced here provides a mechanism for
adding a second binding to an existing resource. There is no difference
between an initial binding added by PUT, COPY, or MKCOL, and additional
bindings added with BIND.
The language here would preclude the future
definition of a DESTROY method which had the semantics of removing the state
of a resource from a server, irregardless of any containment relationships
that may hold it. Such a method could be quite useful for records management
functionality, in order to implement a records disposition policy that
specified deletion at a certain time. My recommended tweak to the language
of section 2 is minor: add the following sentence to the end of the
paragraph:
"It is permissible, however, for future method definitions (e.g., a DESTROY
method) to have semantics that remove all bindings and/or immediately
reclaim system resources."
Agreed to add statement about methods that explicitly have that semantics.
It would be very undesirable if one binding could be destroyed as a side
effect of operating on the resource through a different binding. In
particular, the removal of one binding to a resource (e.g. with a DELETE or
a MOVE) MUST NOT disrupt another binding to that resource, e.g. by turning
that binding into a dangling path segment. The server MUST NOT reclaim
system resources after removing one binding, while other bindings to the
resource remain. In other words, the server MUST maintain the integrity of
a binding. It is permissible, however, for future method definitions (e.g., a
DESTROY method) to have semantics that explicitly remove all bindings
and/or immediately reclaim system resources.
I think it would be more clear to separate out the discussion
of loops and bindings, and make it a separate section (say, 2.2) This issue
comes up frequently enough that it would be good to make it easy to find
this issue in the TOC. Also, a mention of the Already Reported status code
would be good to have here, since it also mentions 506.
Agreed to move 1st paragraph into separate subsection.
Bindings to collections can result in loops, which servers MUST detect when
processing "Depth: infinity" requests. It is sometimes possible to complete
an operation in spite of the presence of a loop. However, the 506 (Loop
Detected) status code is defined in for use in contexts where an
operation is terminated because a loop was encountered.
Creating a new binding to a collection makes each resource associated with
a binding in that collection accessible via a new URI, and thus creates new
URI mappings to those resources but no new bindings.
For example, suppose a new binding CollY is created for collection C1 in the
figure below. It immediately becomes possible to access resource R1 using
the URI /CollY/x.gif and to access resource R2 using the URI /CollY/y.jpg,
but no new bindings for these child resources were created. This is because
bindings are part of the state of a collection, and associate a URI that
is relative to that collection with its target resource. No change to the
bindings in Collection C1 is needed to make its children accessible using
/CollY/x.gif and /CollY/y.jpg.
+-------------------------+
| Root Collection |
| bindings: |
| CollX CollY |
+-------------------------+
| /
| /
| /
+------------------+
| Collection C1 |
| bindings: |
| x.gif y.jpg |
+------------------+
| \
| \
| \
+-------------+ +-------------+
| Resource R1 | | Resource R2 |
+-------------+ +-------------+
Bindings to collections can result in loops, which servers MUST detect when
processing "Depth: infinity" requests. It is sometimes possible to complete
an operation in spite of the presence of a loop. For instance, a PROPFIND
can still succeed if the server uses the new status code
208 (Already Reported) defined in .
However, the 506 (Loop Detected) status code is defined in for use in contexts where an
operation is terminated because a loop was encountered.
...The other is the semantics of the lock operation in the presence
of loopback bindings. I think the handling of If headers is relatively
straightforward. The semantics of locking are not....
After some discussion, the working group agreed that we don't want to define
special semantics for depth:infinity locks, thus the standard lock
sematics apply (see http://lists.w3.org/Archives/Public/w3c-dist-auth/2004OctDec/0271.html).
Suppose a binding from "Binding-Name" to resource R is to be added to a
collection, C. Then if C-MAP is the set of URIs that were mapped to C
before the BIND request, then for each URI "C-URI" in C-MAP, the URI
"C-URI/Binding-Name" is mapped to resource R following the BIND request.
For example, if a binding from "foo.html" to R is added to a collection C,
and if the following URIs are mapped to C:
http://www.example.com/A/1/
http://example.com/A/one/
then the following new mappings to R are introduced:
http://www.example.com/A/1/foo.html
http://example.com/A/one/foo.html
Note that if R is a collection, additional URI mappings are created to the
descendents of R. Also, note that if a binding is made in collection C to C
itself (or to a parent of C), an infinite number of mappings are introduced.
For example, if a binding from "myself" to C is then added to C, the
following infinite number of additional mappings to C are introduced:
http://www.example.com/A/1/myself
http://www.example.com/A/1/myself/myself
...
and the following infinite number of additional mappings to R are introduced:
http://www.example.com/A/1/myself/foo.html
http://www.example.com/A/1/myself/myself/foo.html
...
This section doesn't clearly address the semantics of COPY
with Depth infinity. My recommendation is to add, after paragraph 3, text
like this:
"As specified in [RFC2518], a COPY with Depth infinity causes the collection
resource to be duplicated, all of its bound children to be duplicated, and
their children's bound children, and so on, to the bottom of the containment
hierarchy. All of the segments of the bindings of the destination collection
are the same as for the destination collection. However, the destination
resource for all bindings in the destination collection are different from
those of the source collection, since all resources have been duplicated,
creating new resources with distinct DAV:resource-id properties."
Example added.
As defined in Section 8.8 of , COPY causes the resource identified
by the Request-URI to be duplicated, and makes the new resource accessible
using the URI specified in the Destination header. Upon successful
completion of a COPY, a new binding is created between the last path
segment of the Destination header, and the destination resource. The new
binding is added to its parent collection, identified by the Destination
header minus its final segment.
The following figure shows an example: Suppose that a COPY is issued to
URI-3 for resource R (which is also mapped to URI-1 and URI-2), with the
Destination header set to URI-X. After successful completion of the COPY
operation, resource R is duplicated to create resource R', and a new binding
has been created which creates at least the URI mapping between URI-X and
the new resource (although other URI mappings may also have been created).
URI-1 URI-2 URI-3 URI-X
| | | |
| | | <---- URI Mappings ----> |
| | | |
+---------------------+ +------------------------+
| Resource R | | Resource R' |
+---------------------+ +------------------------+
It might be thought that a COPY request with "Depth: 0" on a collection would
duplicate its bindings, since bindings are part of the collection's state.
This is not the case, however. The definition of Depth in makes it
clear that a "Depth: 0" request does not apply to a collection's members.
Consequently, a COPY with "Depth: 0" does not duplicate the bindings
contained by the collection.
If a COPY request causes an existing resource to be updated, the bindings to
that resource MUST be unaffected by the COPY request. Using the preceding
example, suppose that a COPY request is issued to URI-X for resource R', with
the Destination header set to URI-2. The content and dead properties of
resource R would be updated to be a copy of those of resource R', but the
mappings from URI-1, URI-2, and URI-3 to resource R remain unaffected.
If because of multiple bindings to a resource, more than one source resource
updates a single destination resource, the order of the updates is server
defined.
There should also be some text addressing COPY depth infinity and loops --
in some instances during a COPY with Depth infinity, the server really wants
to recreate the binding that causes the loop, rather than continuing to make
duplicate resources. This is somewhat addressed by the final paragraph in
Section 2.3, but not exactly.
Can be closed after copy/depth:infinity example was added (see
http://lists.w3.org/Archives/Public/w3c-dist-auth/2004OctDec/0181.html).
If a COPY request would cause a new resource to be created as a copy of an
existing resource, and that COPY request has already created a copy of that
existing resource, the COPY request instead creates another binding to the
previous copy, instead of creating a new resource.
As an example of how COPY with Depth infinity would work in the presence of
bindings, consider the following collection:
+------------------+
| Root Collection |
| bindings: |
| CollX |
+------------------+
|
|
+-------------------------------+
| Collection C1 |<-------+
| bindings: | |
| x.gif CollY | |
+-------------------------------+ |
| \ (creates loop) |
| \ |
+-------------+ +------------------+ |
| Resource R1 | | Collection C2 | |
+-------------+ | bindings: | |
| y.gif CollZ | |
+------------------+ |
| | |
| +--------+
|
+-------------+
| Resource R2 |
+-------------+
If a COPY with Depth inifinity is submitted to /CollX, with destination of
/CollA, the outcome of the copy operation is:
+------------------+
| Root Collection |
| bindings: |
| CollX CollA |
+------------------+
| |
| +---------------------------+
| |
+-------------------+ |
| Collection C1 |<------------------+ |
| bindings: | | |
| x.gif CollY | | |
+-------------------+ | |
| \ (creates loop) | |
| \ | |
+-------------+ +-----------------+ | |
| Resource R1 | | Collection C2 | | |
+-------------+ | bindings: | | |
| y.gif CollZ | | |
+-----------------+ | |
| | | |
| +-------+ |
| |
+-------------+ |
| Resource R2 | |
+-------------+ |
|
+-------------------------------+
|
+-------------------+
| Collection C3 |<------------------+
| bindings: | |
| x.gif CollY | |
+-------------------+ |
| \ (creates loop) |
| \ |
+-------------+ +-----------------+ |
| Resource R3 | | Collection C4 | |
+-------------+ | bindings: | |
| y.gif CollZ | |
+-----------------+ |
| | |
| +-------+
|
+-------------+
| Resource R4 |
+-------------+
It might make sense to create an example covering the situation described
in the final paragraph of Section 2.3. I'm not 100% sure I know what
scenario this paragraph addresses, other reading the spec. for the first
time would presumably have a tougher time.
Example added.
Given the following collection hierarchy:
+------------------+
| Root Collection |
| bindings: |
| CollX |
+------------------+
|
|
|
+----------------+
| Collection C1 |
| bindings: |
| x.gif y.gif |
+----------------+
| |
| |
+-------------+
| Resource R1 |
+-------------+
A COPY of /CollX with Depth infinity to /CollY results in the following
collection hierarchy:
+------------------+
| Root Collection |
| bindings: |
| CollX CollY |
+------------------+
| \
| \
| \
+----------------+ +-----------------+
| Collection C1 | | Collection C2 |
| bindings: | | bindings: |
| x.gif y.gif | | x.gif y.gif |
+----------------+ +-----------------+
| | | |
| | | |
+-------------+ +-------------+
| Resource R1 | | Resource R2 |
+-------------+ +-------------+
When there are multiple bindings to a resource, a DELETE applied to that
resource MUST NOT remove any bindings to that resource other than the one
identified by the Request-URI. For example, suppose the collection
identified by the URI "/a" has a binding named "x" to a resource R, and
another collection identified by "/b" has a binding named "y" to the same
resource R. Then a DELETE applied to "/a/x" removes the binding named "x"
from "/a" but MUST NOT remove the binding named "y" from "/b" (i.e. after the
DELETE, "/y/b" continues to identify the resource R). In particular,
although Section 8.6.1 of states that during DELETE processing,
a server "MUST remove any URI for the resource identified by the
Request-URI from collections which contain it as a member", a server that
supports the binding protocol MUST NOT follow this requirement.
When DELETE is applied to a collection, it MUST NOT modify the membership
of any other collection that is not itself a member of the collection being
deleted. For example, if both "/a/.../x" and "/b/.../y" identify the same
collection, C, then applying DELETE to "/a" MUST NOTmust not delete an internal
member from C or from any other collection that is a member of C, because
that would modify the membership of "/b".
If a collection supports the UNBIND method (see ), a DELETE of an
internal member of a collection MAY be implemented as an UNBIND request.
In this case, applying DELETE to a Request-URI has the effect of removing
the binding identified by the final segment of the Request-URI from the
collection identified by the Request-URI minus its final segment. Although
allows a DELETE to be a non-atomic operation, when the DELETE
operation is implemented as an UNBIND, the operation is atomic. In
particular, a DELETE on a hierarchy of resources is simply the removal of a
binding to the collection identified by the Request-URI.
When MOVE is applied to a resource, the other bindings to that resource
MUST be unaffected, and if the resource being moved is a collection, the
bindings to any members of that collection MUST be unaffected. Also, if MOVE
is used with Overwrite:T to delete an existing resource, the constraints
specified for DELETE apply.
If the destination collection of a MOVE request supports the REBIND method
(see ), a MOVE of a resource into that collection MAY be implemented
as a REBIND request. Although allows a MOVE to be a non-atomic
operation, when the MOVE operation is implemented as a REBIND, the operation
is atomic. In particular, applying a MOVE to a Request-URI and a Destination
URI has the effect of removing a binding to a resource (at the Request-URI),
and creating a new binding to that resource (at the Destination URI).
Even when the
Request-URI identifies a collection, the MOVE operation involves only
removing one binding to that collection and adding another.
As an example, suppose that a MOVE is issued to URI-3 for resource R below
(which is also mapped to URI-1 and URI-2), with the Destination header set
to URI-X. After successful completion of the MOVE operation, a new binding
has been created which creates the URI mapping between URI-X and resource R. The
binding corresponding to the final segment of URI-3 has been removed,
which also causes the URI mapping between URI-3 and R to be removed. If
resource R were a collection, old URI-3 based mappings to members of R
would have been removed, and new URI-X based mappings to members of R would
have been created.
>> Before Request:
URI-1 URI-2 URI-3
| | |
| | | <---- URI Mappings
| | |
+---------------------+
| Resource R |
+---------------------+
>> After Request:
URI-1 URI-2 URI-X
| | |
| | | <---- URI Mappings
| | |
+---------------------+
| Resource R |
+---------------------+
It is useful to have some way of determining whether two bindings are to the
same resource. Two resources might have identical contents and properties,
but not be the same resource (e.g. an update to one resource does not affect
the other resource).
The REQUIRED DAV:resource-id property defined in is a resource
identifier, which MUST be unique across all resources for all time. If the
values of DAV:resource-id returned by PROPFIND requests through two bindings
are identical character by character, the client can be assured that the two bindings are to the
same resource.
The DAV:resource-id property is created, and its value assigned, when the
resource is created. The value of DAV:resource-id MUST NOT be changed.
Even after the resource is no longer accessible through any URI, that value
MUST NOT be reassigned to another resource's DAV:resource-id property.
There needs to be some discussion on the interactions of
DAV:resource-id and versioning. As near as I can tell, the intent is that
every revision will have a unique DAV:resource-id value.
Mention in an example.
Any method that creates a new resource MUST assign a new, unique value to
its DAV:resource-id property. For example, a PUT or a COPY that creates a
new resource must assign a new, unique value to the DAV:resource-id property
of that new resource. For example, a PUT applied to a null resource,
COPY (when not overwriting an existing target) and
CHECKIN (see , section 4.4) must assign a new, unique value to the DAV:resource-id property
of the new resource they create.
Change "must not" to
"MUST NOT" (and eliminate the "For example" at the start of the sentence --
perhaps change to "Specifically,"
Fix language, replace MOVE by REBIND (because MOVE may be implemented as COPY/DELETE).
Unclear whether we need more changes.
On the other hand, any method that affects an existing resource MUST NOTmust not
change the value of its DAV:resource-id property. For exampleSpecifically, a PUT or a
COPY that updates an existing resource must not change the value of its
DAV:resource-id property. A MOVEREBIND, since it does not create a new resource,
but only changes the location of an existing resource, must not change the
value of the DAV:resource-id property.
An OPTIONAL DAV:parent-set property on a resource provides a list of the
bindings that associate a collection and a URI segment with that resource. If
the DAV:parent-set property exists on a given resource, it MUST contain
a complete list of all bindings to that resource that the client is
authorized to see. When deciding whether to support the DAV:parent-set
property, server implementers / administrators should balance the benefits
it provides against the cost of maintaining the property and the security
risks enumerated in Sections 8.4 and 8.5 and .
The bind feature introduces the following properties for a resource.
A DAV:allprop PROPFIND request SHOULD NOT return any of the properties
defined by this document. This allows a binding server to perform
efficiently when a naive client, which does not understand the cost of
asking a server to compute all possible live properties, issues a DAV:allprop
PROPFIND request.
The DAV:resource-id property is a REQUIRED property that enables clients to
determine whether two bindings are to the same resource. The value of
DAV:resource-id is a URI, and may use any registered URI scheme that
guarantees the uniqueness of the value across all resources for all time
(e.g. the opaquelocktoken: scheme defined in ).
<!ELEMENT resource-id (href)>
I think it would be helpful to have an example of this
property. I'd be happy to help develop such an example.
Example added, including diagram.
The DAV:parent-set property is an OPTIONAL property that enables clients to
discover what collections contain a binding to this resource (i.e. what
collections have that resource as an internal member). It contains an of
href/segment pair for each collection that has a binding to the resource.
The href identifies the collection, and the segment identifies the binding
name of that resource in that collection.
A given collection MUST appear only once in the DAV:parent-set for any given
binding, even if there are multiple URI mappings to that collection. For
example, if collection C1 is mapped to both /CollX and /CollY, and C1
contains a binding named "x.gif" to a resource R1, then either
[/CollX, x.gif] or [/CollY, x.gif] can appear in the DAV:parent-set of R1,
but not both. But if C1 also had a binding named "y.gif" to R1, then there
would be two entries for C1 in the DAV:binding-set of R1 (i.e. either both
[/CollX, x.gif] and [/CollX, y.gif] or, alternatively, both [/CollY, x.gif]
and [/CollY, y.gif]).
<!ELEMENT parent-set (parent)*>
<!ELEMENT parent (href, segment)>
<!ELEMENT segment (#PCDATA)>
<!-- PCDATA value: segment, as defined in section 3.3 of
-->
For example, if collection C1 is mapped to both /CollX and /CollY, and C1
contains a binding named "x.gif" to a resource R1, then either
[/CollX, x.gif] or [/CollY, x.gif] can appear in the DAV:parent-set of R1,
but not both. But if C1 also had a binding named "y.gif" to R1, then there
would be two entries for C1 in the DAV:binding-set of R1 (i.e. either both
[/CollX, x.gif] and [/CollX, y.gif] or, alternatively, both [/CollY, x.gif]
and [/CollY, y.gif]).
+-------------------------+
| Root Collection |
| bindings: |
| CollX CollY |
+-------------------------+
| /
| /
| /
+-----------------+
| Collection C1 |
| bindings: |
| x.gif y.gif |
+-----------------+
| |
| |
| |
+--------------+
| Resource R1 |
+--------------+
In this case, one possible value for DAV:parent-set property on
"/CollX/x.gif" would be:
<parent-set xmlns="DAV:">
<parent>
<href>/CollX</href>
<segment>x.gif</segment>
</parent>
<parent>
<href>/CollX</href>
<segment>y.gif</segment>
</parent>
</parent-set>
The intent of the BIND method is for its behavior to be atomic.
However, this is never actually stated explicitly in the specification.
Seems like it should be.
Agreed. Steal text from RFC3253 (applies to all method definitions).
The BIND method modifies the collection identified by the Request-URI, by
adding a new binding from the segment specified in the BIND body to the
resource identified in the BIND body.
If a server cannot guarantee the integrity of the binding, the BIND request
MUST fail. Note that it is especially difficult to maintain the integrity of
cross-server bindings. Unless the server where the resource resides knows
about all bindings on all servers to that resource, it may unwittingly
destroy the resource or make it inaccessible without notifying another
server that manages a binding to the resource. For example, if server A
permits creation of a binding to a resource on server B, server A must
notify server B about its binding and must have an agreement with B that B
will not destroy the resource while A's binding exists. Otherwise server B
may receive a DELETE request that it thinks removes the last binding to the
resource and destroy the resource while A's binding still exists.
The precondition DAV:cross-server-binding is defined below
for cases where servers fail cross-server BIND requests because they cannot
guarantee the integrity of cross-server bindings.
By default, if there already is a binding for the specified segment in the
collection, the new binding replaces the existing binding. This default
binding replacement behavior can be overridden using the Overwrite header
defined in Section 9.6 of .
If a BIND request fails, the server state preceding the request MUST be restored.
This method is unsafe and idempotent (see , section 9.1).
Marshalling:
The request MAY include an Overwrite header.The request body MUST be a DAV:bind XML element.
<!ELEMENT bind (segment, href)>
If the request succeeds, the server MUST return 201 (Created) when a new
binding was created and 200 (OK) when an existing binding was replaced.
If a response body for a successful request is included, it MUST be a
DAV:bind-response XML element. Note that this document does not define
any elements for the BIND response body, but the DAV:bind-response
element is defined to ensure interoperability between future extensions
that do define elements for the BIND response body.
<!ELEMENT bind-response ANY>
Preconditions:
(DAV:bind-into-collection): The Request-URI MUST identify a collection.
(DAV:bind-source-exists): The DAV:href element MUST identify a resource.
(DAV:binding-allowed): The resource identified by the DAV:href supports
multiple bindings to it.
(DAV:cross-server-binding): If the resource identified by the DAV:href
element in the request body is on another server from the collection
identified by the Request-URI, the server MUST support cross-server
bindings.
(DAV:name-allowed): The name specified by the DAV:segment is available
for use as a new binding name.
(DAV:can-overwrite): If the collection already contains a binding with
the specified path segment, and if an Overwrite header is included, the
value of the Overwrite header MUST be "T".
(DAV:cycle-allowed): If the DAV:href element identifies a collection,
and if the Request-URI identifies a collection that is a member of that
collection, the server MUST support cycles in the URI namespace.
(DAV:locked-update-allowed): If the collection identified by the
Request-URI is write-locked, then the appropriate token MUST be specified
in an If request header.
(DAV:locked-overwrite-allowed): If the collection already contains a
binding with the specified path segment, and if that binding is protected
by a write-lock, then the appropriate token MUST be specified in an If
request header.
Postconditions:
(DAV:new-binding): The collection MUST have a binding that maps the
segment specified in the DAV:segment element in the request body, to the
resource identified by the DAV:href element in the request body.
>> Request:
BIND /CollY HTTP/1.1
Host: www.example.com
Content-Type: text/xml; charset="utf-8"
Content-Length: xxx
<?xml version="1.0" encoding="utf-8" ?>
<D:bind xmlns:D="DAV:">
<D:segment>bar.html</D:segment>
<D:href>http://www.example.com/CollX/foo.html</D:href>
</D:bind>
>> Response:
HTTP/1.1 201 Created
The server added a new binding to the collection,
"http://www.example.com/CollY", associating "bar.html" with the resource
identified by the URI "http://www.example.com/CollX/foo.html". Clients can
now use the URI "http://www.example.com/CollY/bar.html", to submit requests
to that resource.
The UNBIND method modifies the collection identified by the Request-URI, by
removing the binding identified by the segment specified in the UNBIND body.
Once a resource is unreachable by any URI mapping, the server MAY reclaim
system resources associated with that resource. If UNBIND removes a binding
to a resource, but there remain URI mappings to that resource, the server
MUST NOT reclaim system resources associated with the resource.
If an UNBIND request fails, the server state preceding the request MUST be restored.
This method is unsafe and idempotent (see , section 9.1).
Marshalling:
The request body MUST be a DAV:unbind XML element.
<!ELEMENT unbind (segment)>
If the request succeeds, the server MUST return 200 (OK) when the
binding was successfully deleted.If a response body for a successful request is included, it MUST be a
DAV:unbind-response XML element. Note that this document does not define
any elements for the UNBIND response body, but the DAV:unbind-response
element is defined to ensure interoperability between future extensions
that do define elements for the UNBIND response body.
<!ELEMENT unbind-response ANY>
Preconditions:
(DAV:unbind-from-collection): The Request-URI MUST identify a collection.
(DAV:unbind-source-exists): The DAV:segment element MUST identify a
binding in the collection identified by the Request-URI.
(DAV:locked-update-allowed): If the collection identified by the
Request-URI is write-locked, then the appropriate token MUST be specified
in the request.
(DAV:protected-url-deletion-allowed): If the binding identified by the
segment is protected by a write-lock, then the appropriate token MUST be
specified in the request.
Postconditions:
(DAV:binding-deleted): The collection MUST NOT have a binding for the
segment specified in the DAV:segment element in the request body.
(DAV:lock-deleted): If the internal member URI of the binding specified
by the Request-URI and the DAV:segment element in the
request body was protected by a write-lock at the time of the request,
that write-lock must have been deleted by the request.
>> Request:
UNBIND /CollX HTTP/1.1
Host: www.example.com
Content-Type: text/xml; charset="utf-8"
Content-Length: xxx
<?xml version="1.0" encoding="utf-8" ?>
<D:unbind xmlns:D="DAV:">
<D:segment>foo.html</D:segment>
</D:unbind>
>> Response:
HTTP/1.1 200 OK
The server removed the binding named "foo.html" from the collection,
"http://www.example.com/CollX". A request to the resource named
"http://www.example.com/CollX/foo.html" will return a 404 (Not Found) response.
I'm reading through the BIND specification, and the description of the
REBIND method's operands is a bit unclear to me. I'm assuming the intent is
similar to BIND and UNBIND, each of which clearly state in the first
sentence what role the Request-URI, segment, and href fields play. In my
reading I just jumped right into the spec. at this method (typical reference
reading pattern), and hence I didn't initially see the similarity with the
BIND and UNBIND method operands.
Agreed and fixed (fixed again after it was broken in -08).
I agree with Lisa that the access control implications of REBIND should be made
explicit.
My suggestion is to add the following language to Section 6.
Change:
"It is effectively an atomic form of a MOVE request."
To:
"It is effectively an atomic form of a MOVE request, and MUST be treated as a
MOVE for the purpose of determining access permissions (see RFC 3744, Appendix B)."
Make that statement, but avoid the unnecessary normative reference to RFC3744.
The REBIND method removes a binding to a resource from the collection
identified by the Request-URIa collection, and adds a binding to that resource into another collectionthe collection identified by the Request-URI. The
request body specifies the binding to be removedadded (segment) and the new binding to be createdold binding to be removed
(href). It is effectively an atomic form of a MOVE request,
and MUST be treated the same way as MOVE for the purpose of determining access permissions.
If a REBIND request fails, the server state preceding the request MUST be restored.
This method is unsafe and idempotent (see , section 9.1).
Marshalling:
The request MAY include an Overwrite header.The request body MUST be a DAV:rebind XML element.
<!ELEMENT rebind (segment, href)>
If the request succeeds, the server MUST return 201 (Created) when a
new binding was created and 200 (OK) when an existing binding was replaced.If a response body for a successful request is included, it MUST be a
DAV:rebind-response XML element. Note that this document does not define
any elements for the REBIND response body, but the DAV:rebind-response
element is defined to ensure interoperability between future extensions
that do define elements for the REBIND response body.
<!ELEMENT rebind-response ANY>
Preconditions:
(DAV:rebind-into-collection): The Request-URI MUST identify a
collection.
(DAV:rebind-source-exists): The DAV:href element MUST identify a
resource.
(DAV:cross-server-binding): If the resource identified by the DAV:href
element in the request body is on another server from the collection
identified by the Request-URI, the server MUST support cross-server bindings.
(DAV:name-allowed): The name specified by the DAV:segment is available
for use as a new binding name.
(DAV:can-overwrite): If the collection already contains a binding with
the specified path segment, and if an Overwrite header is included, the
value of the Overwrite header MUST be "T".
(DAV:cycle-allowed): If the DAV:href element identifies a collection,
and if the Request-URI identifies a collection that is a member of that
collection, the server MUST support cycles in the URI namespace.
(DAV:locked-update-allowed): If the collection identified by the
Request-URI is write-locked, then the appropriate token MUST be specified
in the request.
(DAV:protected-url-modification-allowed): If the collection identified
by the Request-URI already contains a binding with the specified path
segment, and if that binding is protected by a write-lock, then the
appropriate token MUST be specified in the request.
(DAV:locked-source-collection-update-allowed): If the collection
identified by the parent collection prefix of the DAV:href URI is
write-locked, then the appropriate token MUST be specified in the request.
(DAV:protected-source-url-deletion-allowed): If the DAV:href URI is
protected by a write lock, then the appropriate token MUST be specified
in the request.
Postconditions:
(DAV:new-binding): The collection MUST have a binding that maps the
segment specified in the DAV:segment element in the request body, to the
resource that was identified by the DAV:href element in the request body.
(DAV:binding-deleted): The URL specified in the DAV:href element in the
request body MUST NOT be mapped to a resource.
(DAV:lock-deleted): If the URL specified in the DAV:href element in the
request body was protected by a write-lock at the time of the request,
that write-lock must have been deleted by the request.
>> Request:
REBIND /CollX HTTP/1.1
Host: www.example.com
Content-Type: text/xml; charset="utf-8"
Content-Length: xxx
<?xml version="1.0" encoding="utf-8" ?>
<D:rebind xmlns:D="DAV:">
<D:segment>foo.html</D:segment>
<D:href>http://www.example.com/CollY/bar.html</D:href>
</D:rebind>
>> Response:
HTTP/1.1 200 OK
The server added a new binding to the collection, "http://www.example.com/CollX",
associating "foo.html" with the resource identified by the URI
"http://www.example.com/CollY/bar.html", and removes the binding named
"bar.html" from the collection identified by the URI "http://www.example.com/CollY". Clients
can now use the URI "http://www.example.com/CollX/foo.html" to submit
requests to that resource, and requests on the URI
"http://www.example.com/CollY/bar.html" will fail with a 404 (Not Found) response.
The 208 (Already Reported) status code can be used inside a DAV:propstat
response element to avoid enumerating the internal members of multiple
bindings to the same collection repeatedly. For each binding to a
collection inside the request's scope, only one will
be reported with a 200 status, while subsequent DAV:response elements for all
other bindings will use the 208 status, and no DAV:response elements for
their descendants are included.
Note that the 208 status will only occur for "Depth: infinity" requests,
and that it is of particular importance when the multiple collection
bindings cause a bind loop as discussed in .
A client can request the DAV:resourceid property in a PROPFIND request to
guarantee that they can accurately reconstruct the binding structure of a
collection with multiple bindings to a single resource.
For backward compatibility with clients not aware of the 208 status code appearing
in multistatus response bodies, it SHOULD NOT be used unless the client
has signalled support for this specification using the "DAV" request
header (see ). Instead, a 506 status should be returned
when a binding loop is discovered.
This allows the server to return the 506 as the top level return status,
if it discovers it before it started the response, or in the middle of a
multistatus, if it discovers it in the middle of streaming out a
multistatus response.
For example, consider a PROPFIND request on /Coll (bound to collection C),
where the members of /Coll are /Coll/Foo (bound to resource R) and /Coll/Bar
(bound to collection C).
>> Request:
PROPFIND /Coll/ HTTP/1.1
Host: www.example.com
Depth: infinity
DAV: bind
Content-Type: text/xml; charset="utf-8"
Content-Length: xxx
<?xml version="1.0" encoding="utf-8" ?>
<D:propfind xmlns:D="DAV:">
<D:prop>
<D:displayname/>
<D:resource-id/>
</D:prop>
</D:propfind>
>> Response:
HTTP/1.1 207 Multi-Status
Content-Type: text/xml; charset="utf-8"
Content-Length: xxx
<?xml version="1.0" encoding="utf-8" ?>
<D:multistatus xmlns:D="DAV:">
<D:response>
<D:href>http://www.example.com/Coll/</D:href>
<D:propstat>
<D:prop>
<D:displayname>Loop Demo</D:displayname>
<D:resource-id>
<D:href
>opaquelocktoken:f81d4fae-7dec-11d0-a765-00a0c91e6bf8</D:href>
</D:resource-id>
</D:prop>
<D:status>HTTP/1.1 200 OK</D:status>
</D:propstat>
</D:response>
<D:response>
<D:href>http://www.example.com/Coll/Foo</D:href>
<D:propstat>
<D:prop>
<D:displayname>Bird Inventory</D:displayname>
<D:resource-id>
<D:href
>opaquelocktoken:f81d4fae-7dec-11d0-a765-00a0c91e6bf9</D:href>
</D:resource-id>
</D:prop>
<D:status>HTTP/1.1 200 OK</D:status>
</D:propstat>
</D:response>
<D:response>
<D:href>http://www.example.com/Coll/Bar</D:href>
<D:propstat>
<D:prop>
<D:displayname>Loop Demo</D:displayname>
<D:resource-id>
<D:href
>opaquelocktoken:f81d4fae-7dec-11d0-a765-00a0c91e6bf8</D:href>
</D:resource-id>
</D:prop>
<D:status>HTTP/1.1 208 Already Reported</D:status>
</D:propstat>
</D:response>
</D:multistatus>
In this example, the client isn't aware of the 208 status code introduced
by this specification. As the "Depth: infinity" PROPFIND request would cause
a loop condition, the whole request is rejected with a 506 status.
>> Request:
PROPFIND /Coll/ HTTP/1.1
Host: www.example.com
Depth: infinity
Content-Type: text/xml; charset="utf-8"
Content-Length: xxx
<?xml version="1.0" encoding="utf-8" ?>
<D:propfind xmlns:D="DAV:">
<D:prop> <D:displayname/> </D:prop>
</D:propfind>
>> Response:
HTTP/1.1 506 Loop Detected
The 506 (Loop Detected) status code indicates that the server terminated an
operation because it encountered an infinite loop while processing a request
with "Depth: infinity". This status indicates that the entire operation
failed.
If the server supports bindings, it MUST return the compliance class name "bind"
as a field in the "DAV" response header (see , section
9.1) from an OPTIONS request on any resource implemented by that
server. A value of "bind" in the "DAV" header MUST indicate that the server
supports all MUST level requirements and REQUIRED features specified in this
document.
This specification introduces the 'DAV' request header that allows clients
to signal compliance to specific WebDAV features. It has the same syntax
as the response header defined in , section
9.1, but MAY be used with any method.
Note that clients MUST NOT submit a specific compliance class name in the
request header unless the specification defining this compliance class
specifically defines its semantics for clients.
Note that if a server chooses to vary the result of a request based on
values in the "DAV" header, the response either MUST NOT be cacheable or
the server MUST mark the response accordingly using the "Vary" header
(see , section 14.44).
Clients SHOULD signal support for all MUST level requirements and REQUIRED
features by submitting a "DAV" request header containing the compliance class
name "bind". In particular, the client MUST understand the 208 status
code defined in .
This section is provided to make WebDAV implementors aware of the security
implications of this protocol.
All of the security considerations of HTTP/1.1 and the WebDAV Distributed
Authoring Protocol specification also apply to this protocol specification.
In addition, bindings introduce several new security concerns and increase
the risk of some existing threats. These issues are detailed below.
In a context where cross-server bindings are supported, creating bindings on
a trusted server may make it possible for a hostile agent to induce users to
send private information to a target on a different server.
Although bind loops were already possible in HTTP 1.1, the introduction
of the BIND method creates a new avenue for clients to create loops
accidentally or maliciously. If the binding and its target are on the same
server, the server may be able to detect BIND requests that would create
loops. Servers are required to detect loops that are caused by bindings to
collections during the processing of any requests with "Depth: infinity".
Denial of service attacks were already possible by posting URIs that were
intended for limited use at heavily used Web sites. The introduction of
BIND creates a new avenue for similar denial of service attacks. If
cross-server bindings are supported, clients can now create bindings at
heavily used sites to target locations that were not designed for heavy usage.
If the DAV:parent-set property is maintained on a resource, the owners of
the bindings risk revealing private locations. The directory structures
where bindings are located are available to anyone who has access to the
DAV:parent-set property on the resource. Moving a binding may reveal its
new location to anyone with access to DAV:parent-set on its resource.
If the server maintains the DAV:parent-set property in response to bindings
created in other administrative domains, it is exposed to hostile attempts to
make it devote resources to adding bindings to the list.
All internationalization considerations mentioned in also apply to
this document.
All IANA considerations mentioned in also apply to this document.
This document is the collaborative product of the authors and Tyson Chihaya,
Jim Davis, Chuck Fay and Judith Slein. This draft has benefited from thoughtful
discussion by Jim Amsden, Peter Carlson, Steve Carter, Ken Coar, Ellis Cohen,
Dan Connolly, Bruce Cragun, Spencer Dawkins, Mark Day, Rajiv Dulepet, David
Durand, Lisa Dusseault, Stefan Eissing, Roy Fielding, Yaron Goland, Joe Hildebrand, Fred Hitt, Alex Hopmann, James Hunt,
Marcus Jager, Chris Kaler, Manoj Kasichainula, Rohit Khare, Brian Korver, Daniel LaLiberte,
Steve Martin, Larry Masinter, Jeff McAffer, Surendra Koduru Reddy, Max Rible,
Sam Ruby, Bradley Sergeant, Nick Shelness, John Stracke, John Tigue, John
Turner, Kevin Wiggen, and other members of the WebDAV working group.
Key words for use in RFCs to Indicate Requirement LevelsHarvard Universitysob@harvard.eduUniform Resource Identifier (URI): Generic SyntaxWorld Wide Web Consortiumtimbl@w3.orgDay Softwarefielding@gbiv.comAdobe Systems IncorporatedLMM@acm.orgHTTP Extensions for Distributed Authoring -- WEBDAVMicrosoft Corporationyarong@microsoft.comDept. Of Information and Computer Science, University of California, Irvineejw@ics.uci.eduNetscapeasad@netscape.comNovellsrcarter@novell.comNovelldcjensen@novell.comHypertext Transfer Protocol -- HTTP/1.1University of California, Irvinefielding@ics.uci.eduW3Cjg@w3.orgCompaq Computer Corporationmogul@wrl.dec.comMIT Laboratory for Computer Sciencefrystyk@w3.orgXerox Corporationmasinter@parc.xerox.comMicrosoft Corporationpaulle@microsoft.comW3Ctimbl@w3.orgExtensible Markup Language (XML) 1.0 (Third Edition)Textuality and Netscapetbray@textuality.comMicrosoftjeanpa@microsoft.comUniversity of Illinois at Chicago and Text Encoding Initiativecmsmcq@uic.eduSun Microsystemseve.maler@east.sun.comfrancois@yergeau.comVersioning Extensions to WebDAV (Web Distributed Authoring and Versioning)Rational Softwaregeoffrey.clemm@rational.comIBMjamsden@us.ibm.comIBMtim_ellison@uk.ibm.comMicrosoftckaler@microsoft.comUC Santa Cruz, Dept. of Computer Scienceejw@cse.ucsc.edu
Add and resolve issues "2.3_COPY_SHARED_BINDINGS" and "2.3_MULTIPLE_COPY".
Add issue "5.1_LOOP_STATUS" and proposed resolution, but keep it open.
Add issues "ED_references" and "4_507_status". Started work on index.
Rename document to "Binding Extensions to Web Distributed Authoring and
Versioning (WebDAV)". Rename "References" to "Normative References".
Close issue "ED_references". Close issue "4_507_status".
Add and close issues "9.2_redirect_loops", "ED_authors" and "ED_updates".
Add section about capability
discovery (DAV header). Close issues "5.1_LOOP_STATUS".
Add and resolve new issue "5.1_506_STATUS_STREAMING".
Update XML spec reference.
Add issue "locking" and resolve as invalid.
Add and close issues "6_precondition_binding_allowed" and
"6_lock_behaviour". Add mailing list and issues list pointers
to front.
Editorial fixes.
Add and resolve issues "1.3_error_negotiation", "2.5_language" and "7.1.1_add_resource_id".
Add historical issue "4_LOCK_BEHAVIOR" and it's resolution for better tracking.
Rewrite Editorial Note. Open and resolve issues "2.6_identical",
"specify_safeness_and_idempotence" and "ED_rfc2026_ref".
Add more index items (no change tracking).
Add and resolve issues "2.3_copy_to_same", "bind_properties", "bind_vs_ACL",
"6_rebind_intro" and "rfc2396bis" (actually an action item).
Fix XML DTD fragment in section 3.3. Make spelling of "Request-URI"
consistent.
Resolved editorial issues raised by Jim Whitehead in .
Add and resolve issues "atomicity", "2_allow_destroy", "2.1_separate_loop_discussion", "2.1.1_bind_loops_vs_locks",
"2.3_copy_depth_infinity", "2.3_copy_example", "2.3_copy_vs_loops", "2.6_resource-id_vs_versions",
"3.2_example" and "6_rebind_premissions".
Add issue "2.6_when_do_ids_change".
Re-open and resolve "6_rebind_intro".