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

RE: [soapbuilders] question re: "typed" nils

Expand Messages
  • Matt Long
    Andrew, point noted, but what if you don t know there is a cupboard in the room? The room contains: empty cupboard empty sideboard empty anything not in the
    Message 1 of 19 , Jun 2, 2001
    • 0 Attachment
      Andrew,

      point noted, but what if you don't know there is a cupboard in the room?
      The room contains:
      empty cupboard
      empty sideboard
      empty anything not in the room

      Is not a bit like answering a "Jeopardy" question(albeit an answer) with
      "What is NOT in my mother's kitchen?" Which is the correct answer to most
      anything (if you know my mother's kitchen).

      is it not true that message intent may be impossible to ensure by omitting
      an element? That is, by omitting the an element the intent may have been
      "nil" or just simply may have been an error.
      ...snatch the pebble from my hand?... ;-) (I'm sure you will!!!!)

      -Matt

      > -----Original Message-----
      > From: Andrew Layman [mailto:yahoo@...]
      > Sent: Friday, June 01, 2001 10:44 PM
      > To: soapbuilders@yahoogroups.com
      > Cc: mrys@...; Allen Brown; ashokma@...
      > Subject: Re: [soapbuilders] question re: "typed" nils
      >
      >
      > If you open an empty cupboard and don't find an elephant
      > inside, are the
      > contents of the cupboard different from opening it and not finding a
      > bicycle? :-)
      >
      > Most database and programming systems do not typically have
      > specific types
      > for different kinds of voids, and for good reason: Something
      > which is not
      > there does not have any properties.
      >
      > This does not mean that there is never a good use for
      > xsi:nil='true'. There
      > may be some times when one wants to indicate not merely the absence of
      > knowledge, but the positive statement that it is known that
      > no value exists.
      > One may even want to add attributes to indicate why no value
      > exists or is
      > known. But, even in these cases, I do not think one would
      > properly use
      > xsi:type.
      >
      > ----- Original Message -----
      > From: "graham glass" <graham-glass@...>
      > To: <soapbuilders@yahoogroups.com>
      > Sent: Friday, June 01, 2001 9:18 AM
      > Subject: [soapbuilders] question re: "typed" nils
      >
      >
      > hi guys,
      >
      > GLUE currently always sends typed data, even when the
      > element is "nil". however, as i design the custom
      > serialization system, i'm finding that this complicates
      > things for developers who want to plug in their own
      > serializers, and so i'm contemplating simplifying their
      > live by dropping the type information for nils.
      >
      > based on previous postings, it sounds like sending a nil
      > in the first place is not as ideal as omitting the element,
      > so i figured that sending a nil without type information
      > would probably not break interop.
      >
      > am i right, or is this a bad move?
      >
      > cheers,
      > graham
      >
      > To unsubscribe from this group, send an email to:
      > soapbuilders-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:
      soapbuilders-unsubscribe@yahoogroups.com



      Your use of Yahoo! Groups is subject to http://docs.yahoo.com/info/terms/
    • Andrew Layman
      Sure, it is possible that a message intent may be impossible to discern without noticing a subtle difference from a rather similar message. But this isn t
      Message 2 of 19 , Jun 2, 2001
      • 0 Attachment
        Sure, it is possible that a message intent may be impossible to discern
        without noticing a subtle difference from a rather similar message. But
        this isn't good practice and should not drive us to design our message
        formats as though this were a typical or important case.

        I guess I was making several points:

        1. It is not good design to create an interface to a service in which
        seemingly subtle differences have significant semantic effects. This
        encourages errors due to human misunderstandings and also due to the fact
        that processes that convert from one representation (e.g. method
        invocations) to another (e.g. SOAP message) may vary just at those subtle
        differences. Further, I think this point is well enough understood that
        programmers typically do not create systems whose proper behavior depends on
        such subtle encoding differences.

        2. Interoperability is achieved by agreeing on the format of messages,
        not method signatures. Different languages have different representations
        and limitations on method signatures, and these differences often do not
        reflect the information that needs to be conveyed but rather some other
        notational characteristics. Various languages have certain restrictions and
        economies, such as positional parameters and allowing omitted parameters
        only at the ends of an argument list. These restrictions do not apply to
        SOAP messages with their named parameters.

        3. "Overloading" a method without changing its name is supported by some
        languages. The most direct representation in a SOAP message pair is to have
        the two messages bear the same element name. However, just as with method
        names that contain characters not legal as XML element names, a more
        sophisticated mapping can avoid the difficulty.

        4. When thinking how to represent information, we need to distinguish
        metaphysical from epistemological issues. By the former I mean "what is
        being claimed about reality" and by the latter "what is claimed about our
        knowledge". For example, there is an important difference between saying
        "Joe has no middle name" and "I don't know whether Joe has a middle name".
        Discussions of NULLs often muddle these two cases together. Michael Rys and
        Allen Brown wrote a paper, which I hope they post here, arguing that xsi:nil
        should be used for the former case and omitted elements for the latter.

        In short, while one could cite a situation in which, in a certain language,
        one could design a pair of overloaded method signatures such that the only
        way to tell them apart was to use xsi:nil in a way that violates the goals
        of point 3 above, this would be unusual, avoidable, and probably bad design.


        ----- Original Message -----
        From: "Matt Long" <mlong@...>
        To: <soapbuilders@yahoogroups.com>
        Sent: Saturday, June 02, 2001 7:01 AM
        Subject: RE: [soapbuilders] question re: "typed" nils


        Andrew,

        point noted, but what if you don't know there is a cupboard in the room?
        The room contains:
        empty cupboard
        empty sideboard
        empty anything not in the room

        Is not a bit like answering a "Jeopardy" question(albeit an answer) with
        "What is NOT in my mother's kitchen?" Which is the correct answer to most
        anything (if you know my mother's kitchen).

        is it not true that message intent may be impossible to ensure by omitting
        an element? That is, by omitting the an element the intent may have been
        "nil" or just simply may have been an error.
        ...snatch the pebble from my hand?... ;-) (I'm sure you will!!!!)

        -Matt

        > -----Original Message-----
        > From: Andrew Layman [mailto:yahoo@...]
        > Sent: Friday, June 01, 2001 10:44 PM
        > To: soapbuilders@yahoogroups.com
        > Cc: mrys@...; Allen Brown; ashokma@...
        > Subject: Re: [soapbuilders] question re: "typed" nils
        >
        >
        > If you open an empty cupboard and don't find an elephant
        > inside, are the
        > contents of the cupboard different from opening it and not finding a
        > bicycle? :-)
        >
        > Most database and programming systems do not typically have
        > specific types
        > for different kinds of voids, and for good reason: Something
        > which is not
        > there does not have any properties.
        >
        > This does not mean that there is never a good use for
        > xsi:nil='true'. There
        > may be some times when one wants to indicate not merely the absence of
        > knowledge, but the positive statement that it is known that
        > no value exists.
        > One may even want to add attributes to indicate why no value
        > exists or is
        > known. But, even in these cases, I do not think one would
        > properly use
        > xsi:type.
        >
        > ----- Original Message -----
        > From: "graham glass" <graham-glass@...>
        > To: <soapbuilders@yahoogroups.com>
        > Sent: Friday, June 01, 2001 9:18 AM
        > Subject: [soapbuilders] question re: "typed" nils
        >
        >
        > hi guys,
        >
        > GLUE currently always sends typed data, even when the
        > element is "nil". however, as i design the custom
        > serialization system, i'm finding that this complicates
        > things for developers who want to plug in their own
        > serializers, and so i'm contemplating simplifying their
        > live by dropping the type information for nils.
        >
        > based on previous postings, it sounds like sending a nil
        > in the first place is not as ideal as omitting the element,
        > so i figured that sending a nil without type information
        > would probably not break interop.
        >
        > am i right, or is this a bad move?
        >
        > cheers,
        > graham
        >
        > To unsubscribe from this group, send an email to:
        > soapbuilders-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:
        soapbuilders-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:
        soapbuilders-unsubscribe@yahoogroups.com



        Your use of Yahoo! Groups is subject to http://docs.yahoo.com/info/terms/
      • James Snell
        ... I find myself in awe of such profound truth. ;-) ... From: Andrew Layman [mailto:yahoo@strongbrains.com] Sent: Friday, June 01, 2001 8:44 PM To:
        Message 3 of 19 , Jun 2, 2001
        • 0 Attachment
          >If you open an empty cupboard and don't find an elephant inside,
          >are the contents of the cupboard different from opening it and
          >not finding a bicycle? :-)

          I find myself in awe of such profound truth. ;-)



          -----Original Message-----
          From: Andrew Layman [mailto:yahoo@...]
          Sent: Friday, June 01, 2001 8:44 PM
          To: soapbuilders@yahoogroups.com
          Cc: mrys@...; Allen Brown; ashokma@...
          Subject: Re: [soapbuilders] question re: "typed" nils


          If you open an empty cupboard and don't find an elephant inside, are the
          contents of the cupboard different from opening it and not finding a
          bicycle? :-)

          Most database and programming systems do not typically have specific
          types for different kinds of voids, and for good reason: Something which
          is not there does not have any properties.

          This does not mean that there is never a good use for xsi:nil='true'.
          There may be some times when one wants to indicate not merely the
          absence of knowledge, but the positive statement that it is known that
          no value exists. One may even want to add attributes to indicate why no
          value exists or is known. But, even in these cases, I do not think one
          would properly use xsi:type.

          ----- Original Message -----
          From: "graham glass" <graham-glass@...>
          To: <soapbuilders@yahoogroups.com>
          Sent: Friday, June 01, 2001 9:18 AM
          Subject: [soapbuilders] question re: "typed" nils


          hi guys,

          GLUE currently always sends typed data, even when the
          element is "nil". however, as i design the custom
          serialization system, i'm finding that this complicates
          things for developers who want to plug in their own serializers, and so
          i'm contemplating simplifying their live by dropping the type
          information for nils.

          based on previous postings, it sounds like sending a nil
          in the first place is not as ideal as omitting the element,
          so i figured that sending a nil without type information
          would probably not break interop.

          am i right, or is this a bad move?

          cheers,
          graham

          To unsubscribe from this group, send an email to:
          soapbuilders-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:
          soapbuilders-unsubscribe@yahoogroups.com



          Your use of Yahoo! Groups is subject to
          http://docs.yahoo.com/info/terms/
        • Andrew Layman
          Thank you. This profound thought occurred to me after days of fasting and meditation on a mountaintop. However, I ve discovered that I was scooped by a few
          Message 4 of 19 , Jun 2, 2001
          • 0 Attachment
            Thank you. This profound thought occurred to me after days of fasting and
            meditation on a mountaintop.

            However, I've discovered that I was scooped by a few millennia:

            "It is impossible for Nothing to be."
            "That which is not, is not, and cannot be or be thought about."
            -- Parmenides, circa 500 BC.

            :-)


            ----- Original Message -----
            From: "James Snell" <jsnell@...>
            To: <soapbuilders@yahoogroups.com>
            Sent: Saturday, June 02, 2001 10:56 AM
            Subject: RE: [soapbuilders] question re: "typed" nils


            >If you open an empty cupboard and don't find an elephant inside,
            >are the contents of the cupboard different from opening it and
            >not finding a bicycle? :-)

            I find myself in awe of such profound truth. ;-)



            -----Original Message-----
            From: Andrew Layman [mailto:yahoo@...]
            Sent: Friday, June 01, 2001 8:44 PM
            To: soapbuilders@yahoogroups.com
            Cc: mrys@...; Allen Brown; ashokma@...
            Subject: Re: [soapbuilders] question re: "typed" nils


            If you open an empty cupboard and don't find an elephant inside, are the
            contents of the cupboard different from opening it and not finding a
            bicycle? :-)

            Most database and programming systems do not typically have specific
            types for different kinds of voids, and for good reason: Something which
            is not there does not have any properties.

            This does not mean that there is never a good use for xsi:nil='true'.
            There may be some times when one wants to indicate not merely the
            absence of knowledge, but the positive statement that it is known that
            no value exists. One may even want to add attributes to indicate why no
            value exists or is known. But, even in these cases, I do not think one
            would properly use xsi:type.

            ----- Original Message -----
            From: "graham glass" <graham-glass@...>
            To: <soapbuilders@yahoogroups.com>
            Sent: Friday, June 01, 2001 9:18 AM
            Subject: [soapbuilders] question re: "typed" nils


            hi guys,

            GLUE currently always sends typed data, even when the
            element is "nil". however, as i design the custom
            serialization system, i'm finding that this complicates
            things for developers who want to plug in their own serializers, and so
            i'm contemplating simplifying their live by dropping the type
            information for nils.

            based on previous postings, it sounds like sending a nil
            in the first place is not as ideal as omitting the element,
            so i figured that sending a nil without type information
            would probably not break interop.

            am i right, or is this a bad move?

            cheers,
            graham

            To unsubscribe from this group, send an email to:
            soapbuilders-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:
            soapbuilders-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:
            soapbuilders-unsubscribe@yahoogroups.com



            Your use of Yahoo! Groups is subject to http://docs.yahoo.com/info/terms/
          • James Snell
            Lot s of good mountain tops around Seattle... Lot s of good places to meditate. ;-) It s interesting however that Parmenides own statement contradicts itself
            Message 5 of 19 , Jun 3, 2001
            • 0 Attachment
              Lot's of good mountain tops around Seattle... Lot's of good places to
              meditate. ;-)

              It's interesting however that Parmenides own statement contradicts
              itself in a way. If Nothing cannot be thought about, how can you state
              that Nothing cannot exist? You must first have a concept of what
              Nothing is before you can postulate as to the actual existence of
              Nothing. Furthermore, if "Nothing" is defined as the total lack of
              anything, then nothing can be nothing more than an idea since there is
              always Something. If Nothing is only an idea, then the very existence
              of the idea of Nothing proves that Nothing exists. Hmmm... I'm going to
              bed now, and will do my best not to dream about "Peri Physeos". ;-)


              -----Original Message-----
              From: Andrew Layman [mailto:yahoo@...]
              Sent: Saturday, June 02, 2001 10:23 PM
              To: soapbuilders@yahoogroups.com
              Subject: Re: [soapbuilders] question re: "typed" nils


              Thank you. This profound thought occurred to me after days of fasting
              and meditation on a mountaintop.

              However, I've discovered that I was scooped by a few millennia:

              "It is impossible for Nothing to be."
              "That which is not, is not, and cannot be or be thought about."
              -- Parmenides, circa 500 BC.

              :-)


              ----- Original Message -----
              From: "James Snell" <jsnell@...>
              To: <soapbuilders@yahoogroups.com>
              Sent: Saturday, June 02, 2001 10:56 AM
              Subject: RE: [soapbuilders] question re: "typed" nils


              >If you open an empty cupboard and don't find an elephant inside, are
              >the contents of the cupboard different from opening it and not finding
              >a bicycle? :-)

              I find myself in awe of such profound truth. ;-)



              -----Original Message-----
              From: Andrew Layman [mailto:yahoo@...]
              Sent: Friday, June 01, 2001 8:44 PM
              To: soapbuilders@yahoogroups.com
              Cc: mrys@...; Allen Brown; ashokma@...
              Subject: Re: [soapbuilders] question re: "typed" nils


              If you open an empty cupboard and don't find an elephant inside, are the
              contents of the cupboard different from opening it and not finding a
              bicycle? :-)

              Most database and programming systems do not typically have specific
              types for different kinds of voids, and for good reason: Something which
              is not there does not have any properties.

              This does not mean that there is never a good use for xsi:nil='true'.
              There may be some times when one wants to indicate not merely the
              absence of knowledge, but the positive statement that it is known that
              no value exists. One may even want to add attributes to indicate why no
              value exists or is known. But, even in these cases, I do not think one
              would properly use xsi:type.

              ----- Original Message -----
              From: "graham glass" <graham-glass@...>
              To: <soapbuilders@yahoogroups.com>
              Sent: Friday, June 01, 2001 9:18 AM
              Subject: [soapbuilders] question re: "typed" nils


              hi guys,

              GLUE currently always sends typed data, even when the
              element is "nil". however, as i design the custom
              serialization system, i'm finding that this complicates
              things for developers who want to plug in their own serializers, and so
              i'm contemplating simplifying their live by dropping the type
              information for nils.

              based on previous postings, it sounds like sending a nil
              in the first place is not as ideal as omitting the element,
              so i figured that sending a nil without type information
              would probably not break interop.

              am i right, or is this a bad move?

              cheers,
              graham

              To unsubscribe from this group, send an email to:
              soapbuilders-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:
              soapbuilders-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:
              soapbuilders-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:
              soapbuilders-unsubscribe@yahoogroups.com



              Your use of Yahoo! Groups is subject to
              http://docs.yahoo.com/info/terms/
            • Andrew Layman
              Well, since you ask... The having the idea of anything in our mind no more proves the existence of that thing than the picture of a man evidences his being in
              Message 6 of 19 , Jun 3, 2001
              • 0 Attachment
                Well, since you ask...

                "The having the idea of anything in our mind no more proves the existence of
                that thing than the picture of a man evidences his being in the world, or
                the visions of a dream make thereby a true history."

                -- John Locke, "An Essay Concerning Human Understanding", 1690.

                Also, from W. T. Jones, 1952, commenting on Parmenides:

                According to Parmenides, "What is not, is not" means that there is no
                nothing,
                that is, that the word "nothing" does not name anything. He seems to have
                reasoned in the following way. You can try to think, for instance, about a
                unicorn;
                if you succeed, you are thinking about something, if only about a fictitious
                animal.
                But now try to think about nothing. You may believe that you are succeeding
                in doing so, but as far as you are thinking at all, you are thinking about
                something
                (that is, there is some object of thought before the mind), not about
                nothing.
                You may call the object about which you are thinking "nothing," but that is
                just a name. There is, and can be, no object, no nothing, named by the name
                "nothing." This apparently is what Parmenides meant when he wrote, "Thou
                canst not know nor utter what is not -- that is impossible."

                That the word "nothing" can appear in a sentence in the grammatical place
                that a noun designating something might also appear is just a quirk of
                language, of no philosophical importance, certainly of no metaphysical
                importance. That this is recognized at some level by everyone is evidenced
                by the Homeric joke of Odysseus's trick on Polyphemus.


                ----- Original Message -----
                From: "James Snell" <jsnell@...>
                To: <soapbuilders@yahoogroups.com>
                Sent: Sunday, June 03, 2001 12:16 AM
                Subject: RE: [soapbuilders] question re: "typed" nils


                Lot's of good mountain tops around Seattle... Lot's of good places to
                meditate. ;-)

                It's interesting however that Parmenides own statement contradicts
                itself in a way. If Nothing cannot be thought about, how can you state
                that Nothing cannot exist? You must first have a concept of what
                Nothing is before you can postulate as to the actual existence of
                Nothing. Furthermore, if "Nothing" is defined as the total lack of
                anything, then nothing can be nothing more than an idea since there is
                always Something. If Nothing is only an idea, then the very existence
                of the idea of Nothing proves that Nothing exists. Hmmm... I'm going to
                bed now, and will do my best not to dream about "Peri Physeos". ;-)


                -----Original Message-----
                From: Andrew Layman [mailto:yahoo@...]
                Sent: Saturday, June 02, 2001 10:23 PM
                To: soapbuilders@yahoogroups.com
                Subject: Re: [soapbuilders] question re: "typed" nils


                Thank you. This profound thought occurred to me after days of fasting
                and meditation on a mountaintop.

                However, I've discovered that I was scooped by a few millennia:

                "It is impossible for Nothing to be."
                "That which is not, is not, and cannot be or be thought about."
                -- Parmenides, circa 500 BC.

                :-)


                ----- Original Message -----
                From: "James Snell" <jsnell@...>
                To: <soapbuilders@yahoogroups.com>
                Sent: Saturday, June 02, 2001 10:56 AM
                Subject: RE: [soapbuilders] question re: "typed" nils


                >If you open an empty cupboard and don't find an elephant inside, are
                >the contents of the cupboard different from opening it and not finding
                >a bicycle? :-)

                I find myself in awe of such profound truth. ;-)



                -----Original Message-----
                From: Andrew Layman [mailto:yahoo@...]
                Sent: Friday, June 01, 2001 8:44 PM
                To: soapbuilders@yahoogroups.com
                Cc: mrys@...; Allen Brown; ashokma@...
                Subject: Re: [soapbuilders] question re: "typed" nils


                If you open an empty cupboard and don't find an elephant inside, are the
                contents of the cupboard different from opening it and not finding a
                bicycle? :-)

                Most database and programming systems do not typically have specific
                types for different kinds of voids, and for good reason: Something which
                is not there does not have any properties.

                This does not mean that there is never a good use for xsi:nil='true'.
                There may be some times when one wants to indicate not merely the
                absence of knowledge, but the positive statement that it is known that
                no value exists. One may even want to add attributes to indicate why no
                value exists or is known. But, even in these cases, I do not think one
                would properly use xsi:type.

                ----- Original Message -----
                From: "graham glass" <graham-glass@...>
                To: <soapbuilders@yahoogroups.com>
                Sent: Friday, June 01, 2001 9:18 AM
                Subject: [soapbuilders] question re: "typed" nils


                hi guys,

                GLUE currently always sends typed data, even when the
                element is "nil". however, as i design the custom
                serialization system, i'm finding that this complicates
                things for developers who want to plug in their own serializers, and so
                i'm contemplating simplifying their live by dropping the type
                information for nils.

                based on previous postings, it sounds like sending a nil
                in the first place is not as ideal as omitting the element,
                so i figured that sending a nil without type information
                would probably not break interop.

                am i right, or is this a bad move?

                cheers,
                graham

                To unsubscribe from this group, send an email to:
                soapbuilders-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:
                soapbuilders-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:
                soapbuilders-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:
                soapbuilders-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:
                soapbuilders-unsubscribe@yahoogroups.com



                Your use of Yahoo! Groups is subject to http://docs.yahoo.com/info/terms/
              • James Snell
                ... Unless, of course, that thing we have the idea of exists only as an idea. Then the very act of thinking it proves its existence -- in fact, demands it. I
                Message 7 of 19 , Jun 4, 2001
                • 0 Attachment
                  >"The having the idea of anything in our mind no more proves the
                  >existence of that thing than the picture of a man evidences his
                  >being in the world, or the visions of a dream make thereby a true
                  >history."

                  Unless, of course, that thing we have the idea of exists only as an
                  idea. Then the very act of thinking it proves its existence -- in fact,
                  demands it.

                  I would agree with Parmenides statement, "Thou canst not know nor utter
                  what is not -- that is impossible." However, you can, through reasoning,
                  have an idea of what is not, or at least what nothing would be if it
                  were possible that nothing could exist. If we talk only in terms of
                  that idea (e.g., the idea of nothing) then Parmenides statement fails to
                  apply.

                  Anyway... I believe we've digressed at bit from the original discussion,
                  no? ;-) .... We'll have to pick this up another time.

                  - James


                  -----Original Message-----
                  From: Andrew Layman [mailto:yahoo@...]
                  Sent: Sunday, June 03, 2001 8:13 PM
                  To: soapbuilders@yahoogroups.com
                  Subject: Re: [soapbuilders] question re: "typed" nils


                  Well, since you ask...

                  "The having the idea of anything in our mind no more proves the
                  existence of that thing than the picture of a man evidences his being in
                  the world, or the visions of a dream make thereby a true history."

                  -- John Locke, "An Essay Concerning Human Understanding", 1690.

                  Also, from W. T. Jones, 1952, commenting on Parmenides:

                  According to Parmenides, "What is not, is not" means that there is no
                  nothing, that is, that the word "nothing" does not name anything. He
                  seems to have reasoned in the following way. You can try to think, for
                  instance, about a unicorn; if you succeed, you are thinking about
                  something, if only about a fictitious animal. But now try to think about
                  nothing. You may believe that you are succeeding in doing so, but as far
                  as you are thinking at all, you are thinking about something (that is,
                  there is some object of thought before the mind), not about nothing. You
                  may call the object about which you are thinking "nothing," but that is
                  just a name. There is, and can be, no object, no nothing, named by the
                  name "nothing." This apparently is what Parmenides meant when he wrote,
                  "Thou canst not know nor utter what is not -- that is impossible."

                  That the word "nothing" can appear in a sentence in the grammatical
                  place that a noun designating something might also appear is just a
                  quirk of language, of no philosophical importance, certainly of no
                  metaphysical importance. That this is recognized at some level by
                  everyone is evidenced by the Homeric joke of Odysseus's trick on
                  Polyphemus.


                  ----- Original Message -----
                  From: "James Snell" <jsnell@...>
                  To: <soapbuilders@yahoogroups.com>
                  Sent: Sunday, June 03, 2001 12:16 AM
                  Subject: RE: [soapbuilders] question re: "typed" nils


                  Lot's of good mountain tops around Seattle... Lot's of good places to
                  meditate. ;-)

                  It's interesting however that Parmenides own statement contradicts
                  itself in a way. If Nothing cannot be thought about, how can you state
                  that Nothing cannot exist? You must first have a concept of what
                  Nothing is before you can postulate as to the actual existence of
                  Nothing. Furthermore, if "Nothing" is defined as the total lack of
                  anything, then nothing can be nothing more than an idea since there is
                  always Something. If Nothing is only an idea, then the very existence
                  of the idea of Nothing proves that Nothing exists. Hmmm... I'm going to
                  bed now, and will do my best not to dream about "Peri Physeos". ;-)


                  -----Original Message-----
                  From: Andrew Layman [mailto:yahoo@...]
                  Sent: Saturday, June 02, 2001 10:23 PM
                  To: soapbuilders@yahoogroups.com
                  Subject: Re: [soapbuilders] question re: "typed" nils


                  Thank you. This profound thought occurred to me after days of fasting
                  and meditation on a mountaintop.

                  However, I've discovered that I was scooped by a few millennia:

                  "It is impossible for Nothing to be."
                  "That which is not, is not, and cannot be or be thought about."
                  -- Parmenides, circa 500 BC.

                  :-)


                  ----- Original Message -----
                  From: "James Snell" <jsnell@...>
                  To: <soapbuilders@yahoogroups.com>
                  Sent: Saturday, June 02, 2001 10:56 AM
                  Subject: RE: [soapbuilders] question re: "typed" nils


                  >If you open an empty cupboard and don't find an elephant inside, are
                  >the contents of the cupboard different from opening it and not finding
                  >a bicycle? :-)

                  I find myself in awe of such profound truth. ;-)



                  -----Original Message-----
                  From: Andrew Layman [mailto:yahoo@...]
                  Sent: Friday, June 01, 2001 8:44 PM
                  To: soapbuilders@yahoogroups.com
                  Cc: mrys@...; Allen Brown; ashokma@...
                  Subject: Re: [soapbuilders] question re: "typed" nils


                  If you open an empty cupboard and don't find an elephant inside, are the
                  contents of the cupboard different from opening it and not finding a
                  bicycle? :-)

                  Most database and programming systems do not typically have specific
                  types for different kinds of voids, and for good reason: Something which
                  is not there does not have any properties.

                  This does not mean that there is never a good use for xsi:nil='true'.
                  There may be some times when one wants to indicate not merely the
                  absence of knowledge, but the positive statement that it is known that
                  no value exists. One may even want to add attributes to indicate why no
                  value exists or is known. But, even in these cases, I do not think one
                  would properly use xsi:type.

                  ----- Original Message -----
                  From: "graham glass" <graham-glass@...>
                  To: <soapbuilders@yahoogroups.com>
                  Sent: Friday, June 01, 2001 9:18 AM
                  Subject: [soapbuilders] question re: "typed" nils


                  hi guys,

                  GLUE currently always sends typed data, even when the
                  element is "nil". however, as i design the custom
                  serialization system, i'm finding that this complicates
                  things for developers who want to plug in their own serializers, and so
                  i'm contemplating simplifying their live by dropping the type
                  information for nils.

                  based on previous postings, it sounds like sending a nil
                  in the first place is not as ideal as omitting the element,
                  so i figured that sending a nil without type information
                  would probably not break interop.

                  am i right, or is this a bad move?

                  cheers,
                  graham

                  To unsubscribe from this group, send an email to:
                  soapbuilders-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:
                  soapbuilders-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:
                  soapbuilders-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:
                  soapbuilders-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:
                  soapbuilders-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:
                  soapbuilders-unsubscribe@yahoogroups.com



                  Your use of Yahoo! Groups is subject to
                  http://docs.yahoo.com/info/terms/
                Your message has been successfully submitted and would be delivered to recipients shortly.