Loading ...
Sorry, an error occurred while loading the content.
 

Re: [rest-discuss] REST's Definition of Resource (fwd)

Expand Messages
  • Sam Hunting
    ... I don t think that the word interpretation was meant as what a software component called an interpreter does . The question is, I think How do the
    Message 1 of 30 , Oct 3, 2003
      > > <p>
      > > Professor <a
      > > href="http://www.gseis.ucla.edu/faculty/pages/svenonius.html">
      > > Elaine Svenonious</a> is the author of several papers, ....
      > > </p>
      > >
      > >

      > > Is REST's interpretation of this snippet that the author uses the
      > > URI to identify a resource that 'is' the person Elaine Svenonious?
      > > And that the author's intention is *not* to identify a resource
      > > that 'is' the Web page.
      >

      > REST is an architectural style, not an interpreter.

      I don't think that the word "interpretation" was meant as "what a software
      component called an interpreter does". The question is, I think "How do
      the architects of REST and users of the REST style interpret". (Since the
      word "interpret" comes along higher up in the thread, I thought I'd better
      clarify.)

      Sam Hunting
      eTopicality, Inc.

      ---------------------------------------------------------------------------
      Co-editor: ISO Reference Model for Topic Maps
      Topic map consulting and training: www.etopicality.com
      Free open source topic map tools: www.gooseworks.org
      XML Topic Maps: Creating and Using Topic Maps for the Web.
      Addison-Wesley, ISBN 0-201-74960-2.
      ---------------------------------------------------------------------------
    • Walden Mathews
      Jan, An opportunity for some extreme, and perhaps useful, irony has presented itself in this last question of yours. (I wonder, though, how much this really
      Message 2 of 30 , Oct 3, 2003
        Jan,

        An opportunity for some extreme, and perhaps useful, irony has
        presented itself in this last question of yours. (I wonder, though, how
        much this really has to do with REST.)

        Did you realize that you used the word "hat" where you probably
        meant to use the word "what"?

        What did you mean to identify with that word (or should I say "hat
        were you trying to identify..."?). Does the context of that sentence
        affect the *architecture* of the English language?

        Walden


        ----- Original Message -----
        From: "Jan Algermissen" <jalgermissen@...>
        To: "Roy T. Fielding" <fielding@...>
        Cc: <algermissen@...>; "rest-discuss" <rest-discuss@yahoogroups.com>
        Sent: Friday, October 03, 2003 6:36 AM
        Subject: Re: [rest-discuss] REST's Definition of Resource (fwd)


        : "Roy T. Fielding" wrote:
        :
        : > We have no way of knowing whether the URI, as used by the author,
        : > is intended to identify the person, their home page, or some other
        : > aspect of past responses that the author simply had no other URI
        : > to reference. That's why it is useful to look at the text inside
        : > the anchor, both here and in other authors' use of the URI, and
        : > any other metadata that might be laying about.
        :
        :
        : Roy, hat is the meaning of 'author' in what you write? Is it
        :
        : - the person that 'issued' the URI in the first place ("naming authority")
        :
        :
        : - or the person that used the URI as the value of the href attribute?
        :
        :
        :
        : Jan
        :
        :
        :
        : > ....Roy
        : >
        : >
        : > To unsubscribe from this group, send an email to:
        : > rest-discuss-unsubscribe@yahoogroups.com
        : >
        : >
        : >
        : > Your use of Yahoo! Groups is subject to
        http://docs.yahoo.com/info/terms/
        :
        : --
        : Jan Algermissen http://www.topicmapping.com
        : Consultant & Programmer http://www.gooseworks.org
        :
        :
        : To unsubscribe from this group, send an email to:
        : rest-discuss-unsubscribe@yahoogroups.com
        :
        :
        :
        : Your use of Yahoo! Groups is subject to http://docs.yahoo.com/info/terms/
        :
        :
        :
      • Jan Algermissen
        ... Yes, I meant what sorry about that. ... I tried, but I admit that I don t get what you are up to. Is this an analogy I don t get? Jan ... -- Jan
        Message 3 of 30 , Oct 3, 2003
          Walden Mathews wrote:
          >
          > Jan,
          >
          > An opportunity for some extreme, and perhaps useful, irony has
          > presented itself in this last question of yours. (I wonder, though, how
          > much this really has to do with REST.)
          >
          > Did you realize that you used the word "hat" where you probably
          > meant to use the word "what"?

          Yes, I meant "what" sorry about that.

          >
          > What did you mean to identify with that word (or should I say "hat
          > were you trying to identify..."?). Does the context of that sentence
          > affect the *architecture* of the English language?

          I tried, but I admit that I don't get what you are up to. Is this an
          analogy I don't get?


          Jan


          >
          > Walden
          >
          > ----- Original Message -----
          > From: "Jan Algermissen" <jalgermissen@...>
          > To: "Roy T. Fielding" <fielding@...>
          > Cc: <algermissen@...>; "rest-discuss" <rest-discuss@yahoogroups.com>
          > Sent: Friday, October 03, 2003 6:36 AM
          > Subject: Re: [rest-discuss] REST's Definition of Resource (fwd)
          >
          > : "Roy T. Fielding" wrote:
          > :
          > : > We have no way of knowing whether the URI, as used by the author,
          > : > is intended to identify the person, their home page, or some other
          > : > aspect of past responses that the author simply had no other URI
          > : > to reference. That's why it is useful to look at the text inside
          > : > the anchor, both here and in other authors' use of the URI, and
          > : > any other metadata that might be laying about.
          > :
          > :
          > : Roy, hat is the meaning of 'author' in what you write? Is it
          > :
          > : - the person that 'issued' the URI in the first place ("naming authority")
          > :
          > :
          > : - or the person that used the URI as the value of the href attribute?
          > :
          > :
          > :
          > : Jan
          > :
          > :
          > :
          > : > ....Roy
          > : >
          > : >
          > : > To unsubscribe from this group, send an email to:
          > : > rest-discuss-unsubscribe@yahoogroups.com
          > : >
          > : >
          > : >
          > : > Your use of Yahoo! Groups is subject to
          > http://docs.yahoo.com/info/terms/
          > :
          > : --
          > : Jan Algermissen http://www.topicmapping.com
          > : Consultant & Programmer http://www.gooseworks.org
          > :
          > :
          > : To unsubscribe from this group, send an email to:
          > : rest-discuss-unsubscribe@yahoogroups.com
          > :
          > :
          > :
          > : Your use of Yahoo! Groups is subject to http://docs.yahoo.com/info/terms/
          > :
          > :
          > :
          >
          >
          > To unsubscribe from this group, send an email to:
          > rest-discuss-unsubscribe@yahoogroups.com
          >
          >
          >
          > Your use of Yahoo! Groups is subject to http://docs.yahoo.com/info/terms/

          --
          Jan Algermissen http://www.topicmapping.com
          Consultant & Programmer http://www.gooseworks.org
        • Sam Hunting
          Walden: ... It s good news that irony is not dead! ... I guess I don t see the relation between a typo ( hat for what ) and questioning what Roy meant by
          Message 4 of 30 , Oct 3, 2003
            Walden:

            Now I am confused enough not to lurk. With much slashing:

            > Walden Mathews wrote:
            > >
            > > Jan,
            > >
            > > An opportunity for some extreme, and perhaps useful, irony has
            > > presented itself in this last question of yours. (I wonder, though, how
            > > much this really has to do with REST.)

            It's good news that irony is not dead!

            > >
            > > Did you realize that you used the word "hat" where you probably
            > > meant to use the word "what"?
            >
            > Yes, I meant "what" sorry about that.
            >
            > >
            > > What did you mean to identify with that word (or should I say "hat
            > > were you trying to identify..."?). Does the context of that sentence
            > > affect the *architecture* of the English language?
            >
            > I tried, but I admit that I don't get what you are up to. Is this an
            > analogy I don't get?

            I guess I don't see the relation between a "typo" ("hat" for "what") and
            questioning what Roy meant by the word "author" is in a sentence.

            That is, by "author" do we mean the "coiner" of the URI, who inserts that
            data into a document, or the "issuing authority" of the URI, that serves
            up a representation of the resource identified by that the URI (in
            context). Both? Some third meanning?

            Walden, are you saying the question is meaningless and cannot be asked, at
            least within the context of a REST architectural discussion?

            That the word "author" is a typo for some other correct word?

            (Thanks to all for a very enlightening discussion.)

            > > : "Roy T. Fielding" wrote:
            > > :
            > > : > We have no way of knowing whether the URI, as used by the author,
            > > : > is intended to identify the person, their home page, or some other
            > > : > aspect of past responses that the author simply had no other URI
            > > : > to reference. That's why it is useful to look at the text inside
            > > : > the anchor, both here and in other authors' use of the URI, and
            > > : > any other metadata that might be laying about.
            > > :
            > > :
            > > : Roy, hat is the meaning of 'author' in what you write? Is it
            > > :
            > > : - the person that 'issued' the URI in the first place ("naming authority")
            > > :
            > > :
            > > : - or the person that used the URI as the value of the href attribute?






            Sam Hunting
            eTopicality, Inc.

            ---------------------------------------------------------------------------
            Co-editor: ISO Reference Model for Topic Maps
            Topic map consulting and training: www.etopicality.com
            Free open source topic map tools: www.gooseworks.org
            XML Topic Maps: Creating and Using Topic Maps for the Web.
            Addison-Wesley, ISBN 0-201-74960-2.
            ---------------------------------------------------------------------------
          • Walden Mathews
            Jan, You were asking What allows this URI, embedded in this context, to identify a person as opposed to a string of bits? [paraphrased]. By analogy, what
            Message 5 of 30 , Oct 3, 2003
              Jan,

              You were asking "What allows this URI, embedded in this context,
              to identify a person as opposed to a string of bits?" [paraphrased].

              By analogy, what allowed the typo "hat" in your previous message to
              identify the concept "what" when I first read it?

              More to the point, what does the above question have to do with
              the architectural style of the system in which the phenomenon occurred?
              I'm not sure I know the answer; if you don't either, then I'd say we're
              stuck.

              Walden

              ----- Original Message -----
              From: "Jan Algermissen" <jalgermissen@...>
              Cc: "rest-discuss" <rest-discuss@yahoogroups.com>
              Sent: Friday, October 03, 2003 2:02 PM
              Subject: Re: [rest-discuss] REST's Definition of Resource (fwd)


              : Walden Mathews wrote:
              : >
              : > Jan,
              : >
              : > An opportunity for some extreme, and perhaps useful, irony has
              : > presented itself in this last question of yours. (I wonder, though, how
              : > much this really has to do with REST.)
              : >
              : > Did you realize that you used the word "hat" where you probably
              : > meant to use the word "what"?
              :
              : Yes, I meant "what" sorry about that.
              :
              : >
              : > What did you mean to identify with that word (or should I say "hat
              : > were you trying to identify..."?). Does the context of that sentence
              : > affect the *architecture* of the English language?
              :
              : I tried, but I admit that I don't get what you are up to. Is this an
              : analogy I don't get?
              :
              :
              : Jan
              :
              :
              : >
              : > Walden
              : >
              : > ----- Original Message -----
              : > From: "Jan Algermissen" <jalgermissen@...>
              : > To: "Roy T. Fielding" <fielding@...>
              : > Cc: <algermissen@...>; "rest-discuss" <rest-discuss@yahoogroups.com>
              : > Sent: Friday, October 03, 2003 6:36 AM
              : > Subject: Re: [rest-discuss] REST's Definition of Resource (fwd)
              : >
              : > : "Roy T. Fielding" wrote:
              : > :
              : > : > We have no way of knowing whether the URI, as used by the author,
              : > : > is intended to identify the person, their home page, or some other
              : > : > aspect of past responses that the author simply had no other URI
              : > : > to reference. That's why it is useful to look at the text inside
              : > : > the anchor, both here and in other authors' use of the URI, and
              : > : > any other metadata that might be laying about.
              : > :
              : > :
              : > : Roy, hat is the meaning of 'author' in what you write? Is it
              : > :
              : > : - the person that 'issued' the URI in the first place ("naming
              authority")
              : > :
              : > :
              : > : - or the person that used the URI as the value of the href attribute?
              : > :
              : > :
              : > :
              : > : Jan
              : > :
              : > :
              : > :
              : > : > ....Roy
              : > : >
              : > : >
              : > : > To unsubscribe from this group, send an email to:
              : > : > rest-discuss-unsubscribe@yahoogroups.com
              : > : >
              : > : >
              : > : >
              : > : > Your use of Yahoo! Groups is subject to
              : > http://docs.yahoo.com/info/terms/
              : > :
              : > : --
              : > : Jan Algermissen http://www.topicmapping.com
              : > : Consultant & Programmer http://www.gooseworks.org
              : > :
              : > :
              : > : To unsubscribe from this group, send an email to:
              : > : rest-discuss-unsubscribe@yahoogroups.com
              : > :
              : > :
              : > :
              : > : Your use of Yahoo! Groups is subject to
              http://docs.yahoo.com/info/terms/
              : > :
              : > :
              : > :
              : >
              : >
              : > To unsubscribe from this group, send an email to:
              : > rest-discuss-unsubscribe@yahoogroups.com
              : >
              : >
              : >
              : > Your use of Yahoo! Groups is subject to
              http://docs.yahoo.com/info/terms/
              :
              : --
              : Jan Algermissen http://www.topicmapping.com
              : Consultant & Programmer http://www.gooseworks.org
              :
              :
              : To unsubscribe from this group, send an email to:
              : rest-discuss-unsubscribe@yahoogroups.com
              :
              :
              :
              : Your use of Yahoo! Groups is subject to http://docs.yahoo.com/info/terms/
              :
              :
              :
            • Walden Mathews
              ... Maybe not; I m wishing now I kept my ironic mouth shut. ... I don t either. Jan was asking what allowed a given URI in a given context to identify a
              Message 6 of 30 , Oct 3, 2003
                :
                : It's good news that irony is not dead!

                Maybe not; I'm wishing now I kept my ironic mouth shut.

                : I guess I don't see the relation between a "typo" ("hat" for "what") and
                : questioning what Roy meant by the word "author" is in a sentence.

                I don't either. Jan was asking what allowed a given URI in a given
                context to identify a *person*, as opposed to a more expected
                consequence, like identifying a document. My point was simply that
                even defective identifiers sometimes work well, provided a supportive
                enough context. You know, it begs the question: What allows identifiers
                to work at all? And then: Is REST doing/promising something other
                than the usual in that department? I don't think so.
                :
                : Walden, are you saying the question is meaningless and cannot be asked, at
                : least within the context of a REST architectural discussion?

                No. BTW, I think when Roy said "author", he was referring to the
                person or agent who composed the hypertext snippet.

                :
                : That the word "author" is a typo for some other correct word?

                No. Hmm, what author word might the author have intended?

                -Walden
              • Jan Algermissen
                Let me try to be as explicit as possible. There are two subjects: A: the abstract concept of the Person Elaine Svenonious B: the document (it s abstract
                Message 7 of 30 , Oct 3, 2003
                  Let me try to be as explicit as possible.

                  There are two subjects:

                  A: the abstract concept of the Person "Elaine Svenonious"
                  B: the 'document' (it's abstract notion, not the bits and bytes of course)

                  There are two Resources representing these subjects:

                  S_A: the resource representing subject A
                  S_B: the resource representing subject B


                  The URI http://www.gseis.ucla.edu/faculty/pages/svenonius.html
                  can be an identifier for either one of the subjects, depending on the
                  URI-users (the one authoring the href attribute) intention.

                  Is that correct?


                  Bottom line: a single URI can identify several resources.

                  Correct?


                  Jan
                • Walden Mathews
                  ... From: Jan Algermissen To: Walden Mathews Cc: Sam Hunting ;
                  Message 8 of 30 , Oct 3, 2003
                    ----- Original Message -----
                    From: "Jan Algermissen" <jalgermissen@...>
                    To: "Walden Mathews" <waldenm@...>
                    Cc: "Sam Hunting" <shunting@...>; "rest-discuss"
                    <rest-discuss@yahoogroups.com>
                    Sent: Friday, October 03, 2003 4:06 PM
                    Subject: Re: [rest-discuss] REST's Definition of Resource (fwd)


                    : Let me try to be as explicit as possible.
                    :
                    : There are two subjects:
                    :
                    : A: the abstract concept of the Person "Elaine Svenonious"
                    : B: the 'document' (it's abstract notion, not the bits and bytes of
                    course)
                    :
                    : There are two Resources representing these subjects:
                    :
                    : S_A: the resource representing subject A
                    : S_B: the resource representing subject B

                    In that case your resources happen to be representations, if
                    I'm to take you literally. While on the other hand you could have
                    the resources *be* these subjects.

                    :
                    :
                    : The URI http://www.gseis.ucla.edu/faculty/pages/svenonius.html
                    : can be an identifier for either one of the subjects, depending on the
                    : URI-users (the one authoring the href attribute) intention.
                    :
                    : Is that correct?

                    Sounds wrong to me. The context an author may create around
                    an identifier (a URI) does not affect the mapping of representations,
                    which is where the consistency needs to be perceived, and hence
                    the notion of identity observed
                    .
                    :
                    :
                    : Bottom line: a single URI can identify several resources.
                    :
                    : Correct?

                    No. The word is not " identify". It might be connote. I'm not
                    a linguist; I don't know.

                    Walden
                  • Roy T. Fielding
                    ... Because REST doesn t interpret hypertext -- people do. ... No. All URIs are names. Some URIs also contain what might be viewed as an address involving
                    Message 9 of 30 , Oct 3, 2003
                      >> Suppose the HTML snippet:
                      >>
                      >> <p>
                      >> Professor <a
                      >> href="http://www.gseis.ucla.edu/faculty/pages/svenonius.html">
                      >> Elaine Svenonious</a> is the author of several papers, ....
                      >> </p>
                      >
                      > Why does REST 'allow' the interpretation that the author uses the URI
                      > to
                      > identify a resource that 'is' the person Elaine Svenonious?

                      Because REST doesn't interpret hypertext -- people do.

                      > The person "Elaine Svenonious" is not directly addressable (in the
                      > sense
                      > that it
                      > does not have a location) and allowing a URI to identify it implies
                      > (IMHO) the
                      > use of URIs as names (which I thought was purely an RDF/Semantic Web
                      > issue).

                      No. All URIs are names. Some URIs also contain what might be viewed as
                      an address involving multiple indirect mappings, and very few URIs
                      contain a directly addressed maping (e.g., file://127.0.0.1/).
                      "http" URIs, in particular, have at least one mandatory indirect
                      mapping, which makes them capable of referencing any concept.

                      > Hmm, propably my non-understanding can also be expressed like this:
                      > How can an architectural style for distributed systems use object
                      > references
                      > (here URIs) as object references AND names for abstract concepts?

                      Because they are not object references -- they reference resources.
                      BTW, it is a style for network-based systems -- "distributed" means
                      something else.

                      > [You just cannot reference an asbtract concept directly, you can only
                      > refer
                      > to it by name.]

                      You cannot access an HTTP-based resource directly, either.

                      ....Roy
                    • Roy T. Fielding
                      ... No. It affects the result of identifying the resource, and that in turn affects how people using that URI perceive it as identifying. But, what it
                      Message 10 of 30 , Oct 3, 2003
                        On Friday, October 3, 2003, at 03:07 AM, Jan Algermissen wrote:
                        > "Roy T. Fielding" wrote:
                        >> We have no way of knowing whether the URI, as used by the author,
                        >> is intended to identify the person, their home page, or some other
                        >> aspect of past responses that the author simply had no other URI
                        >> to reference. That's why it is useful to look at the text inside
                        >> the anchor, both here and in other authors' use of the URI, and
                        >> any other metadata that might be laying about.
                        >
                        > So it is true that an author's use of a URI provides a context that
                        > affects what resource the URI is actually intended to identify, yes?

                        No. It affects the result of identifying the resource, and that in turn
                        affects how people using that URI perceive it as identifying. But, what
                        it actually identifies is unknown to the link author, unless they happen
                        to know by some other method (e.g., being the owner, asking the owner,
                        etc.). People generally just follow the leader and assume that what
                        one person thinks it identifies is what they think it identifies.

                        > So, a single URI can in fact identify several different resources
                        > (depending on the URI-user's intention). Did I get that right?

                        A single URI can be used to indirectly identify multiple resources,
                        even though it only directly identifies whatever the maintainer of
                        that mapping wishes to maintain.

                        By author I meant "link maker", not resource owner.

                        ....Roy
                      • Jan Algermissen
                        ... Roy, what exactly do you mean by indirectly identify ? Given the definition of resource beeing a mapping from URI to a set of representations I read
                        Message 11 of 30 , Oct 4, 2003
                          Jan wrote:

                          > > So, a single URI can in fact identify several different resources
                          > > (depending on the URI-user's intention). Did I get that right?

                          Roy wrote:

                          > A single URI can be used to indirectly identify multiple resources,
                          > even though it only directly identifies whatever the maintainer of
                          > that mapping wishes to maintain.

                          Roy, what exactly do you mean by "indirectly identify"?




                          Given the definition of 'resource' beeing a mapping from URI to a set of
                          representations I read your answer to mean that


                          A single URI can be used to indirectly identify multiple mappings
                          from that URI to sets of representations.


                          Did I get that right?





                          Appologies for being so difficult, but I seem to be missing a critical
                          piece of the whole idea.


                          Jan




                          --
                          Jan Algermissen http://www.topicmapping.com
                          Consultant & Programmer http://www.gooseworks.org
                        • Roy T. Fielding
                          Please look up indirect in a dictionary. That is what I mean. I hate ontology questions from SW land -- they are a total waste of time. ....Roy
                          Message 12 of 30 , Oct 4, 2003
                            Please look up indirect in a dictionary. That is what I mean.
                            I hate ontology questions from SW land -- they are a total
                            waste of time.

                            ....Roy
                          • Walden Mathews
                            Jan, ... I just want to correct one seemingly trivial mistake in the above: the resource is not a mapping from URI to representations. If it were, then you
                            Message 13 of 30 , Oct 4, 2003
                              Jan,
                              :
                              : Given the definition of 'resource' beeing a mapping from URI to a set of
                              : representations I read your answer to mean that ...
                              :

                              I just want to correct one seemingly trivial mistake in the above: the
                              resource is not a mapping from URI to representations. If it were, then
                              you could only have one URI identifying a resource. The mapping
                              is from time to representations. That's a little harder to grasp, but it
                              looks like this:

                              TIME REPRESENTATION
                              t1 r1
                              t1 r2
                              t2 r1
                              t2 r2
                              t2 r3
                              t3 (null)
                              t4 r2

                              etc.

                              What you are seeing above is a "resource". You can mint as many
                              URI as you please to name it.

                              Formally:
                              sig Resource {map: Time->Representation}
                              sig URI {identifies: Resource}

                              E.g.: Two URI on the same Resource:
                              u1, u2: URI | u1.identifies = u2.identifies
                              => "every t->r tuple found vi u1 is also found via u2,
                              and vice versa"

                              Walden
                            • Walden Mathews
                              Roy, I m not sure what an ontology question is, but I was wondering whether *you meant* by a URI can identify many resources indirectly that a. (my hunch)
                              Message 14 of 30 , Oct 4, 2003
                                Roy,

                                I'm not sure what an ontology question is, but I was wondering
                                whether *you meant* by "a URI can identify many resources
                                indirectly" that

                                a. (my hunch) that a URI may identify a resource which references
                                other resources by embedding their URI in its represnetations,
                                to arbitrarily many recurses of this indirection, or

                                b. (probably Jan's assumption) that a URI may 'identify' a resource
                                by inference, more like connotation that identification, which I
                                believe is a trap I stepped into yesterday in trying to navigate
                                this territory.

                                Would you mind just clarifying that?

                                Thanks,
                                Walden

                                ----- Original Message -----
                                From: "Roy T. Fielding" <fielding@...>
                                To: <algermissen@...>
                                Cc: "rest-discuss" <rest-discuss@yahoogroups.com>
                                Sent: Saturday, October 04, 2003 4:48 AM
                                Subject: Re: [rest-discuss] REST's Definition of Resource (fwd)


                                : Please look up indirect in a dictionary. That is what I mean.
                                : I hate ontology questions from SW land -- they are a total
                                : waste of time.
                                :
                                : ....Roy
                                :
                                :
                                :
                                : To unsubscribe from this group, send an email to:
                                : rest-discuss-unsubscribe@yahoogroups.com
                                :
                                :
                                :
                                : Your use of Yahoo! Groups is subject to http://docs.yahoo.com/info/terms/
                                :
                                :
                                :
                              • Jan Algermissen
                                ... I completely agree. I tried to keep the meaning stuff out of this discussion as much as I could - sorry if that wasn t clear. I am asking all these
                                Message 15 of 30 , Oct 4, 2003
                                  "Roy T. Fielding" wrote:
                                  >
                                  > Please look up indirect in a dictionary. That is what I mean.
                                  > I hate ontology questions from SW land -- they are a total
                                  > waste of time.

                                  I completely agree. I tried to keep the 'meaning' stuff out of this
                                  discussion as much as I could - sorry if that wasn't clear.

                                  I am asking all these question for pure practical reasons (see below).

                                  What I did not understand was that on the one hand REST says that the
                                  naming authority decides what the resource is that a given URI identifies
                                  and that on the other hand people on this list said that the user of
                                  a URI (the link maker) decides what resource it identifies.

                                  Then I understood you to say that

                                  - a URI (directly) identifies the one resource that the naming authority
                                  decides it to identify.

                                  - a URI (may) **indirectly** identify other resources depending on the
                                  intention of the URI-user (the link-maker)


                                  I haven't seen the idea of URIs indirectly identifying resources used in
                                  the context of REST before, that's why I asked what you (or REST) mean by
                                  it.


                                  The practical problem I have is this:

                                  The moment a user agent sends a request to the origin server, any 'indirection
                                  information' is lost due to REST's unique addressing context. Certainly,
                                  the origin server 'knows' what resource the URI (directly) identifies because
                                  that is controlled by the naming authority. But all intermediaries have no way
                                  of seeing/accessing any information about indirection in identification.

                                  I read about the idea (thanks Mark!) that intermediaries can provide
                                  information about the resources that the URIs identify they see in requests
                                  'running through' them and weave that information into the representation
                                  returned to the user agent. It was then that I understood the advantages of
                                  a unique addressing context, because the intermediary knows that a given
                                  URI unambiguously identifies a single resource.

                                  Given the idea of 'indirect identification' the question arises, how
                                  intermediaries can usefully enhance representations with information they
                                  obtained, for example, from an RDF document.

                                  I like the collaboration idea, which seems to come down to the possibility
                                  that many *indirect* identifications can affect/alter a naming authority's
                                  decision about the *direct* identification. Wonder how intermediaries can
                                  contribute to this process.


                                  Jan



                                  > ....Roy
                                  >
                                  >
                                  > To unsubscribe from this group, send an email to:
                                  > rest-discuss-unsubscribe@yahoogroups.com
                                  >
                                  >
                                  >
                                  > Your use of Yahoo! Groups is subject to http://docs.yahoo.com/info/terms/

                                  --
                                  Jan Algermissen http://www.topicmapping.com
                                  Consultant & Programmer http://www.gooseworks.org
                                • Roy T. Fielding
                                  ... It would be better to say that the link-maker is indirectly identifying other resources (for which he is ignorant of a better URI, if any) through use of a
                                  Message 16 of 30 , Oct 4, 2003
                                    > What I did not understand was that on the one hand REST says that the
                                    > naming authority decides what the resource is that a given URI
                                    > identifies
                                    > and that on the other hand people on this list said that the user of
                                    > a URI (the link maker) decides what resource it identifies.
                                    >
                                    > Then I understood you to say that
                                    >
                                    > - a URI (directly) identifies the one resource that the naming
                                    > authority
                                    > decides it to identify.
                                    >
                                    > - a URI (may) **indirectly** identify other resources depending on the
                                    > intention of the URI-user (the link-maker)

                                    It would be better to say that the link-maker is indirectly identifying
                                    other resources (for which he is ignorant of a better URI, if any)
                                    through use of a URI that seems to result in what he wants when it
                                    is used. Whether or not his intended resource is, in fact, the one
                                    identified by the URI (by means of the owner maintaining the mapping)
                                    is unknown to that link-maker.

                                    It gets worse: over time, if the majority of link-makers have a
                                    different
                                    opinion about the nature of the resource being identified than the
                                    owner,
                                    they might change the owner's opinion as well. [oh, you mention that
                                    later]

                                    > I haven't seen the idea of URIs indirectly identifying resources used
                                    > in
                                    > the context of REST before, that's why I asked what you (or REST) mean
                                    > by
                                    > it.

                                    Hmm, well, from my point of view, the idea that "http" URIs are ever
                                    used to directly identify anything is a bit absurd. The interface is
                                    there to hide the nature of what is behind it, and it cannot do that
                                    very well if the clients have to know what it is.

                                    > The practical problem I have is this:
                                    >
                                    > The moment a user agent sends a request to the origin server, any
                                    > 'indirection
                                    > information' is lost due to REST's unique addressing context.
                                    > Certainly,
                                    > the origin server 'knows' what resource the URI (directly) identifies
                                    > because
                                    > that is controlled by the naming authority. But all intermediaries
                                    > have no way
                                    > of seeing/accessing any information about indirection in
                                    > identification.

                                    The server doesn't know either -- it relies on the human maintainer to
                                    know
                                    well enough to set up the correct mechanism for the mapping.

                                    > I read about the idea (thanks Mark!) that intermediaries can provide
                                    > information about the resources that the URIs identify they see in
                                    > requests
                                    > 'running through' them and weave that information into the
                                    > representation
                                    > returned to the user agent. It was then that I understood the
                                    > advantages of
                                    > a unique addressing context, because the intermediary knows that a
                                    > given
                                    > URI unambiguously identifies a single resource.
                                    >
                                    > Given the idea of 'indirect identification' the question arises, how
                                    > intermediaries can usefully enhance representations with information
                                    > they
                                    > obtained, for example, from an RDF document.

                                    They can only state what they know, and that very carefully. Just like
                                    real life, an intermediary must either state only what it has observed
                                    in the past, or state its predictions on the future in terms that make
                                    clear they are only guesses based on past observations.

                                    > I like the collaboration idea, which seems to come down to the
                                    > possibility
                                    > that many *indirect* identifications can affect/alter a naming
                                    > authority's
                                    > decision about the *direct* identification. Wonder how intermediaries
                                    > can
                                    > contribute to this process.

                                    That is one of the main weaknesses of HTTP -- there is no way for a
                                    proxy to distinguish between authoritative responses and ones that
                                    it cooked up on its own.

                                    I'll be in England this week, so no more from me on this thread.

                                    ....Roy
                                  • Chuck Hinson
                                    ... c. a URI may identify a resource which makes use of other resources to produce its representation. --Chuck
                                    Message 17 of 30 , Oct 4, 2003
                                      Walden Mathews wrote:

                                      >Roy,
                                      >
                                      >I'm not sure what an ontology question is, but I was wondering
                                      >whether *you meant* by "a URI can identify many resources
                                      >indirectly" that
                                      >
                                      >a. (my hunch) that a URI may identify a resource which references
                                      > other resources by embedding their URI in its represnetations,
                                      > to arbitrarily many recurses of this indirection, or
                                      >
                                      >b. (probably Jan's assumption) that a URI may 'identify' a resource
                                      > by inference, more like connotation that identification, which I
                                      > believe is a trap I stepped into yesterday in trying to navigate
                                      > this territory.
                                      >
                                      >
                                      c. a URI may identify a resource which makes use of other resources to
                                      produce its representation.


                                      --Chuck
                                    • Jan Algermissen
                                      ... Hmm, this suggests (IMHO) that it is bad practice to use URIs to identify other resources than the ones they identify by definition their naming
                                      Message 18 of 30 , Oct 4, 2003
                                        "Roy T. Fielding" wrote:
                                        >
                                        > > What I did not understand was that on the one hand REST says that the
                                        > > naming authority decides what the resource is that a given URI
                                        > > identifies
                                        > > and that on the other hand people on this list said that the user of
                                        > > a URI (the link maker) decides what resource it identifies.
                                        > >
                                        > > Then I understood you to say that
                                        > >
                                        > > - a URI (directly) identifies the one resource that the naming
                                        > > authority
                                        > > decides it to identify.
                                        > >
                                        > > - a URI (may) **indirectly** identify other resources depending on the
                                        > > intention of the URI-user (the link-maker)
                                        >
                                        > It would be better to say that the link-maker is indirectly identifying
                                        > other resources (for which he is ignorant of a better URI, if any)
                                        > through use of a URI that seems to result in what he wants when it
                                        > is used. Whether or not his intended resource is, in fact, the one
                                        > identified by the URI (by means of the owner maintaining the mapping)
                                        > is unknown to that link-maker.
                                        >
                                        > It gets worse: over time, if the majority of link-makers have a
                                        > different
                                        > opinion about the nature of the resource being identified than the
                                        > owner,
                                        > they might change the owner's opinion as well. [oh, you mention that
                                        > later]

                                        Hmm, this suggests (IMHO) that it is 'bad practice' to use URIs to
                                        identify other resources than the ones they identify by definition their
                                        naming authority.

                                        Example: Given a URI that identifies a resource that 'is a document' (the
                                        abstraction of a document), say a paper about presidential democracy, it
                                        would be bad practice to use that URI to identify the idea 'presidetion
                                        democracy'.

                                        I think this is what I don't get in my head. It is exactly what people do
                                        all the time but this common practice seems to be very unRESTful.

                                        Or am I just confused by RDF/SW abuses of the Web?

                                        Taken to extreme: if thousands of RDF authors would use my email address to
                                        identify 'me' in RDF documents, should I then call my ISP (the naming authority)
                                        and have him change the definition of what resource my email address identifies?


                                        However, thanks for all the answers to myquestions. You all have been
                                        very helpful.

                                        Jan



                                        --
                                        Jan Algermissen http://www.topicmapping.com
                                        Consultant & Programmer http://www.gooseworks.org
                                      • Walden Mathews
                                        Chuck, Could you elaborate, and especially draw a distinction between your c and my a ? Thanks, Walden ... From: Chuck Hinson To:
                                        Message 19 of 30 , Oct 4, 2003
                                          Chuck,

                                          Could you elaborate, and especially draw a distinction between
                                          your "c" and my "a"?

                                          Thanks,
                                          Walden

                                          ----- Original Message -----
                                          From: "Chuck Hinson" <cmhinson@...>
                                          To: "Walden Mathews" <waldenm@...>; "rest-discuss"
                                          <rest-discuss@yahoogroups.com>
                                          Sent: Saturday, October 04, 2003 9:58 AM
                                          Subject: Re: [rest-discuss] REST's Definition of Resource (fwd)


                                          :
                                          : Walden Mathews wrote:
                                          :
                                          : >Roy,
                                          : >
                                          : >I'm not sure what an ontology question is, but I was wondering
                                          : >whether *you meant* by "a URI can identify many resources
                                          : >indirectly" that
                                          : >
                                          : >a. (my hunch) that a URI may identify a resource which references
                                          : > other resources by embedding their URI in its represnetations,
                                          : > to arbitrarily many recurses of this indirection, or
                                          : >
                                          : >b. (probably Jan's assumption) that a URI may 'identify' a resource
                                          : > by inference, more like connotation that identification, which I
                                          : > believe is a trap I stepped into yesterday in trying to navigate
                                          : > this territory.
                                          : >
                                          : >
                                          : c. a URI may identify a resource which makes use of other resources to
                                          : produce its representation.
                                          :
                                          :
                                          : --Chuck
                                          :
                                          :
                                          :
                                        • Chuck Hinson
                                          ... see inline... ... If I understand you correctly -- Directly identified resource: http://example.com/story Indirectly identified resource:
                                          Message 20 of 30 , Oct 4, 2003
                                            Walden Mathews wrote:

                                            >Chuck,
                                            >
                                            >Could you elaborate, and especially draw a distinction between
                                            >your "c" and my "a"?
                                            >
                                            >
                                            see inline...

                                            >Thanks,
                                            >Walden
                                            >
                                            >----- Original Message -----
                                            >From: "Chuck Hinson" <cmhinson@...>
                                            >To: "Walden Mathews" <waldenm@...>; "rest-discuss"
                                            ><rest-discuss@yahoogroups.com>
                                            >Sent: Saturday, October 04, 2003 9:58 AM
                                            >Subject: Re: [rest-discuss] REST's Definition of Resource (fwd)
                                            >
                                            >
                                            >:
                                            >: Walden Mathews wrote:
                                            >:
                                            >: >Roy,
                                            >: >
                                            >: >I'm not sure what an ontology question is, but I was wondering
                                            >: >whether *you meant* by "a URI can identify many resources
                                            >: >indirectly" that
                                            >: >
                                            >: >a. (my hunch) that a URI may identify a resource which references
                                            >: > other resources by embedding their URI in its represnetations,
                                            >: > to arbitrarily many recurses of this indirection, or
                                            >
                                            If I understand you correctly --
                                            Directly identified resource: http://example.com/story
                                            Indirectly identified resource: http://example.com/pictures/picture1
                                            where the directly identified resource has a representation that is an
                                            HTML document and that HTML document includes the tag <img
                                            src="http://example.com/pictures/picture1">


                                            >: >
                                            >: >b. (probably Jan's assumption) that a URI may 'identify' a resource
                                            >: > by inference, more like connotation that identification, which I
                                            >: > believe is a trap I stepped into yesterday in trying to navigate
                                            >: > this territory.
                                            >: >
                                            >: >
                                            >: c. a URI may identify a resource which makes use of other resources to
                                            >: produce its representation.
                                            >:
                                            >

                                            Directly identified resource: http://example.com/story
                                            Indirectly identified resource: file://docroot/story/story.html
                                            where a representation of the indirectly identified resource is used to
                                            produce a representation of the directly identified resource. This is
                                            the case for what is commonly referred to as 'static content'.

                                            Or
                                            Directly identified resource: http://example.com/pictures/picture1
                                            Indirectly identified resources: file://docroot/story/images/picture1.jpg,
                                            file://docroot/story/images/picture1.png,
                                            file://docroot/story/images/picture1.gif
                                            where content negotation on the directly identified resource results in
                                            the use of the representation of one of the indirectly identified
                                            resources as the representation of the directly identified resource.

                                            --Chuck
                                          • Walden Mathews
                                            ... Yes, that is what I meant. ... Hold it. You re calling an URI a resource above. That s going to cause confusion. Are you saying that the first URI
                                            Message 21 of 30 , Oct 4, 2003
                                              : >: >a. (my hunch) that a URI may identify a resource which references
                                              : >: > other resources by embedding their URI in its represnetations,
                                              : >: > to arbitrarily many recurses of this indirection, or
                                              : >
                                              : If I understand you correctly --
                                              : Directly identified resource: http://example.com/story
                                              : Indirectly identified resource: http://example.com/pictures/picture1
                                              : where the directly identified resource has a representation that is an
                                              : HTML document and that HTML document includes the tag <img
                                              : src="http://example.com/pictures/picture1">

                                              Yes, that is what I meant.

                                              :
                                              :
                                              : >: >
                                              : >: >b. (probably Jan's assumption) that a URI may 'identify' a resource
                                              : >: > by inference, more like connotation that identification, which I
                                              : >: > believe is a trap I stepped into yesterday in trying to navigate
                                              : >: > this territory.
                                              : >: >
                                              : >: >
                                              : >: c. a URI may identify a resource which makes use of other resources to
                                              : >: produce its representation.
                                              : >:
                                              : >
                                              :
                                              : Directly identified resource: http://example.com/story
                                              : Indirectly identified resource: file://docroot/story/story.html
                                              : where a representation of the indirectly identified resource is used to
                                              : produce a representation of the directly identified resource. This is
                                              : the case for what is commonly referred to as 'static content'.

                                              Hold it. You're calling an URI a resource above. That's going to
                                              cause confusion. Are you saying that the first URI identifies two
                                              resources, namely the one that it identifies directly, and also the one
                                              that can be identified directly by the second URI? First let me get
                                              that part straight.

                                              If that's your claim, then I'll now need to know more about what
                                              the phrase "used to produce a representation of" really means. Do
                                              you mean they share a representation? Do you mean something
                                              which occurs in the server implementation that cannot be observed
                                              on the network? Are you sure this is any form of identification at
                                              all, let alone "indirect identification"? I have no idea, as the phrase
                                              "indirect identification" is new to me.
                                              :
                                              : Or
                                              : Directly identified resource: http://example.com/pictures/picture1
                                              : Indirectly identified resources: file://docroot/story/images/picture1.jpg,
                                              : file://docroot/story/images/picture1.png,
                                              : file://docroot/story/images/picture1.gif
                                              : where content negotation on the directly identified resource results in
                                              : the use of the representation of one of the indirectly identified
                                              : resources as the representation of the directly identified resource.

                                              That doesn't really seem any different, except that you've introduced
                                              content negotiation. Are you implying that all content negotiation
                                              scenarios involve the "indirect identification" phenomenon?


                                              I'm concerned about the quality of the terminology, especially the
                                              term "indirect". To me, it doesn't mean indeterminate; it just means
                                              that the identification is multi-stage. In
                                              this setting, that implies identifier that points to identifier that
                                              points...
                                              until something other than a pointer results.

                                              If the phenomenon is something other than that, I'll bet a better
                                              terminology can be found. Like "selection via association" maybe.

                                              Walden
                                            • Chuck Hinson
                                              ... Arrrggg. I did not call a URI a resource. I identified a resource by using its name. I refuse to go around talking about resources as the resource
                                              Message 22 of 30 , Oct 4, 2003
                                                Walden Mathews wrote:

                                                >: >: c. a URI may identify a resource which makes use of other resources to
                                                >: >: produce its representation.
                                                >: >:
                                                >: >
                                                >:
                                                >: Directly identified resource: http://example.com/story
                                                >: Indirectly identified resource: file://docroot/story/story.html
                                                >: where a representation of the indirectly identified resource is used to
                                                >: produce a representation of the directly identified resource. This is
                                                >: the case for what is commonly referred to as 'static content'.
                                                >
                                                >Hold it. You're calling an URI a resource above. That's going to
                                                >cause confusion. Are you saying that the first URI identifies two
                                                >resources, namely the one that it identifies directly, and also the one
                                                >that can be identified directly by the second URI? First let me get
                                                >that part straight.
                                                >
                                                >
                                                Arrrggg. I did not call a URI a resource. I identified a resource by
                                                using its name.
                                                I refuse to go around talking about resources as 'the resource
                                                identified by http://...'. It's a resource and it has a name - when I
                                                talk about it, I'll use its name. Do you refer to other people as 'the
                                                human identified by Fred', or do you just say Fred.

                                                (Yes, I'm frustrated. I'm tired of having to argure the whole way back
                                                to first principles every time a statement is made. Sometimes I think
                                                some wicked grammar school teacher has exacted her revenge by having
                                                taught me an alternate version of the English language - one that seems
                                                similar on the surface, but is different enough to cause me problems any
                                                time I try to convey ideas to someone else. Either that, or I'm some
                                                kind of strange Italian who can't communicate without using a whiteboard.)

                                                >If that's your claim, then I'll now need to know more about what
                                                >the phrase "used to produce a representation of" really means. Do
                                                >you mean they share a representation? Do you mean something
                                                >which occurs in the server implementation that cannot be observed
                                                >on the network? Are you sure this is any form of identification at
                                                >all, let alone "indirect identification"? I have no idea, as the phrase
                                                >"indirect identification" is new to me.
                                                >

                                                When a representation is produced (in order to satisfy a GET, for
                                                instance), the bits that comprise it are either dynamically calculated
                                                at the time the request is made, or retrieved from some other resource(s).
                                                In the cases where those bits are retrieved from some other resource, I
                                                claim that those other resources are indirectly indentified.

                                                Now perhaps I've twisted the definition of 'indirect', but I don't think
                                                so.

                                                If I identify a group of people as 'White males', do I not also also
                                                indirectly indentify myself by virtue of belonging to that group.


                                                >:
                                                >: Or
                                                >: Directly identified resource: http://example.com/pictures/picture1
                                                >: Indirectly identified resources: file://docroot/story/images/picture1.jpg,
                                                >: file://docroot/story/images/picture1.png,
                                                >: file://docroot/story/images/picture1.gif
                                                >: where content negotation on the directly identified resource results in
                                                >: the use of the representation of one of the indirectly identified
                                                >: resources as the representation of the directly identified resource.
                                                >
                                                >That doesn't really seem any different, except that you've introduced
                                                >content negotiation. Are you implying that all content negotiation
                                                >scenarios involve the "indirect identification" phenomenon?
                                                >
                                                >
                                                >
                                                I didn't intend to imply that its true for ALL content negotiation
                                                (although it might very well be). I just saw content negotiation for
                                                static content as a way to illustrate how directly identifiying one
                                                resource MIGHT result in the indirect identification of other resources.

                                                >I'm concerned about the quality of the terminology, especially the
                                                >term "indirect". To me, it doesn't mean indeterminate; it just means
                                                >that the identification is multi-stage. In
                                                >this setting, that implies identifier that points to identifier that
                                                >points...
                                                >until something other than a pointer results.
                                                >
                                                I don't understand why you think it means indeterminate. It's no more
                                                indeterminate than the relationship between a resource and its
                                                representations over time.

                                                >
                                                >If the phenomenon is something other than that, I'll bet a better
                                                >terminology can be found. Like "selection via association" maybe.
                                                >
                                                >
                                                I dont understand what "selection via association" means.

                                                In my mind, indirect indentification is, in part, a result of resource
                                                composition - resources can be composed of other resources.

                                                --Chuck

                                                PS. Have you ever had the experience where you've said a word over and
                                                over so many times that it eventually begins to sound like gibberish.
                                                Well, when I pronounce the sequence of letters i-d-e-n-t-i-f-i-e-d, I
                                                hear only noise.
                                              • Walden Mathews
                                                ... Chuck (does that answer your last question?), ordinarily I d agree with you, but you re talking about a system now in which a URI identifies more than one
                                                Message 23 of 30 , Oct 4, 2003
                                                  : >: Directly identified resource: http://example.com/story
                                                  : >: Indirectly identified resource: file://docroot/story/story.html
                                                  : >: where a representation of the indirectly identified resource is used to
                                                  : >: produce a representation of the directly identified resource. This is
                                                  : >: the case for what is commonly referred to as 'static content'.
                                                  : >
                                                  : >Hold it. You're calling an URI a resource above. That's going to
                                                  : >cause confusion. Are you saying that the first URI identifies two
                                                  : >resources, namely the one that it identifies directly, and also the one
                                                  : >that can be identified directly by the second URI? First let me get
                                                  : >that part straight.
                                                  : >
                                                  : >
                                                  : Arrrggg. I did not call a URI a resource. I identified a resource by
                                                  : using its name.
                                                  : I refuse to go around talking about resources as 'the resource
                                                  : identified by http://...'. It's a resource and it has a name - when I
                                                  : talk about it, I'll use its name. Do you refer to other people as 'the
                                                  : human identified by Fred', or do you just say Fred.

                                                  Chuck (does that answer your last question?), ordinarily I'd agree
                                                  with you, but you're talking about a system now in which a URI
                                                  identifies more than one resource, and that caused me a problem
                                                  in trying to understand what you wrote. So sorry.

                                                  :
                                                  : (Yes, I'm frustrated. I'm tired of having to argure the whole way back
                                                  : to first principles every time a statement is made. Sometimes I think
                                                  : some wicked grammar school teacher has exacted her revenge by having
                                                  : taught me an alternate version of the English language - one that seems
                                                  : similar on the surface, but is different enough to cause me problems any
                                                  : time I try to convey ideas to someone else. Either that, or I'm some
                                                  : kind of strange Italian who can't communicate without using a whiteboard.)
                                                  :
                                                  : >If that's your claim, then I'll now need to know more about what
                                                  : >the phrase "used to produce a representation of" really means. Do
                                                  : >you mean they share a representation? Do you mean something
                                                  : >which occurs in the server implementation that cannot be observed
                                                  : >on the network? Are you sure this is any form of identification at
                                                  : >all, let alone "indirect identification"? I have no idea, as the phrase
                                                  : >"indirect identification" is new to me.
                                                  : >
                                                  :
                                                  : When a representation is produced (in order to satisfy a GET, for
                                                  : instance), the bits that comprise it are either dynamically calculated
                                                  : at the time the request is made, or retrieved from some other resource(s).
                                                  : In the cases where those bits are retrieved from some other resource, I
                                                  : claim that those other resources are indirectly indentified.

                                                  I don't follow the language there, sorry. You could always argue
                                                  that the representation produced is "retrieved from some other resource",
                                                  whether it was calculated dynamically or not. I'm looking for significant
                                                  distinctions to understand "indirect identification" and not finding them.
                                                  That's my frustration with this thread.

                                                  :
                                                  : Now perhaps I've twisted the definition of 'indirect', but I don't think
                                                  : so.
                                                  :
                                                  : If I identify a group of people as 'White males', do I not also also
                                                  : indirectly indentify myself by virtue of belonging to that group.

                                                  You're definitely twisting the definition of "identify" here, because
                                                  identifying that group *does not* identify you, in the sense that it does
                                                  not make you *definitively recognizable* among the group. It's not
                                                  that it's indirect (i.e., via some intermediate means), it's that it's in-
                                                  determinate: "Hey, there's a group of white guys. Which one is Chuck?"

                                                  :
                                                  :
                                                  : >:
                                                  : >: Or
                                                  : >: Directly identified resource: http://example.com/pictures/picture1
                                                  : >: Indirectly identified resources:
                                                  file://docroot/story/images/picture1.jpg,
                                                  : >: file://docroot/story/images/picture1.png,
                                                  : >: file://docroot/story/images/picture1.gif
                                                  : >: where content negotation on the directly identified resource results in
                                                  : >: the use of the representation of one of the indirectly identified
                                                  : >: resources as the representation of the directly identified resource.
                                                  : >
                                                  : >That doesn't really seem any different, except that you've introduced
                                                  : >content negotiation. Are you implying that all content negotiation
                                                  : >scenarios involve the "indirect identification" phenomenon?
                                                  : >
                                                  : >
                                                  : >
                                                  : I didn't intend to imply that its true for ALL content negotiation
                                                  : (although it might very well be). I just saw content negotiation for
                                                  : static content as a way to illustrate how directly identifiying one
                                                  : resource MIGHT result in the indirect identification of other resources.

                                                  That's interesting, because I think it's true that identifying a resource
                                                  is a means of indirectly identifying a *representation* (in the sense that
                                                  one and only one will be dished up per GET [identified], and the
                                                  representation is
                                                  not the thing you directly identified, but was available through an
                                                  intermediate mapping [indirect]). It's also true that REST does not
                                                  guarantee
                                                  deterministic identification of a representation via its resource: if there
                                                  are two, you might get either one.

                                                  :
                                                  : >I'm concerned about the quality of the terminology, especially the
                                                  : >term "indirect". To me, it doesn't mean indeterminate; it just means
                                                  : >that the identification is multi-stage. In
                                                  : >this setting, that implies identifier that points to identifier that
                                                  : >points...
                                                  : >until something other than a pointer results.
                                                  : >
                                                  : I don't understand why you think it means indeterminate. It's no more
                                                  : indeterminate than the relationship between a resource and its
                                                  : representations over time.

                                                  I was guessing; indirection does lay some of the groundwork for
                                                  possible indeterminism. It's not like they're totally unrelated.
                                                  :
                                                  : >
                                                  : >If the phenomenon is something other than that, I'll bet a better
                                                  : >terminology can be found. Like "selection via association" maybe.
                                                  : >
                                                  : >
                                                  : I dont understand what "selection via association" means.

                                                  Well, I think Roy was explaining that sometimes hypetext authors
                                                  make reference to a resource that's not quite what they mean, but
                                                  it's associated with the intended meaning, and it's close in meaning,
                                                  and it's close at hand. So they select it. That would be selection
                                                  by association, wouldn't it?

                                                  :
                                                  : In my mind, indirect indentification is, in part, a result of resource
                                                  : composition - resources can be composed of other resources.

                                                  What an interesting statement! What would that look like? I
                                                  visualize it as the union of two or more resource mappings. How
                                                  about you?

                                                  :
                                                  : --Chuck
                                                  :
                                                  : PS. Have you ever had the experience where you've said a word over and
                                                  : over so many times that it eventually begins to sound like gibberish.
                                                  : Well, when I pronounce the sequence of letters i-d-e-n-t-i-f-i-e-d, I
                                                  : hear only noise.

                                                  When I do it, I hear a sequence of letters. Are you pronouncing the
                                                  dashes, by any chance? :-)

                                                  Walden
                                                • Chuck Hinson
                                                  ... Hmmm. It s suddenly become clear to me that I ve been going on about something entirely different (I think) than what Roy was talking about. My apologies
                                                  Message 24 of 30 , Oct 4, 2003
                                                    Walden Mathews wrote:

                                                    >: >
                                                    >: >If the phenomenon is something other than that, I'll bet a better
                                                    >: >terminology can be found. Like "selection via association" maybe.
                                                    >: >
                                                    >: >
                                                    >: I dont understand what "selection via association" means.
                                                    >
                                                    >Well, I think Roy was explaining that sometimes hypetext authors
                                                    >make reference to a resource that's not quite what they mean, but
                                                    >it's associated with the intended meaning, and it's close in meaning,
                                                    >and it's close at hand. So they select it. That would be selection
                                                    >by association, wouldn't it?
                                                    >
                                                    >
                                                    Hmmm. It's suddenly become clear to me that I've been going on about
                                                    something entirely different (I think) than what Roy was talking about.
                                                    My apologies for the noise.

                                                    --Chuck
                                                  Your message has been successfully submitted and would be delivered to recipients shortly.