IPP Mail Archive: RE: IPP> MED - Proposed final ABNF for Med

RE: IPP> MED - Proposed final ABNF for Media Size Self Describing Names

From: Bergman, Ron (Ron.Bergman@Hitachi-hkis.com)
Date: Wed Jul 11 2001 - 18:42:47 EDT

  • Next message: Hastings, Tom N: "IPP> RE: IFX> ippget Spec Change Request"

    Tom,

    As I said in my last email, I believe that separating the registration
    requirements from the media size name ABNF is cleaner. It is then very
    obvious that the class-xx name is restricted to the given names. Addition
    of reg-class-xx-name to the media size names ABNF muddies the water. When
    this is only in a section that describes the requirements for new class
    names, everything is clean. The best way to define that the parser must be
    able to deal with new names is with text, not the ABNF.

    I don't see anyone else commenting on this issue. Does that mean that no
    one cares?

            Ron

    -----Original Message-----
    From: Hastings, Tom N [mailto:hastings@cp10.es.xerox.com]
    Sent: Wednesday, July 11, 2001 12:19 PM
    To: Bergman, Ron
    Cc: ipp (E-mail)
    Subject: RE: IPP> MED - Proposed final ABNF for Media Size Self
    Describing Names

    So to be clear about the issue about how to represent the ABNF for future
    registered class names:

    (a) in the ABNF in the Media Standardized Name standard itself and
    (b) the ABNF in the flat file on the PWG server

    (a) For the ABNF in the standard to which sending implementations and
    receiving implementations are to conform, should the ABNF include the
    reg-class-in-name and reg-class-mm-name productions or not in the single
    ABNF? If the productions for the registered class names is removed from the
    single ABNF, then they will appear later in the document explaining what the
    syntax is for additional registered class names (as in the D0.9 draft).

    So should the first productions in the standard be:

       media-size-self-describing-name =
          ( class-in "_" size-name "_" short-dim "x" long-dim "in" ) |
          ( class-mm "_" size-name "_" short-dim "x" long-dim "mm" )

       class-in = "na" | "asme" | "oe" | "custom" | reg-class-in-name
       class-mm = "iso" | "jis" | "jpn" | "prc" | "roc" | "om" | "custom" |
                  reg-class-mm-name

       reg-class-in-name = ( lowalpha | digit ) *( lowalpha | digit | "." )
       reg-class-mm-name = ( lowalpha | digit ) *( lowalpha | digit | "." )

    or:

       media-size-self-describing-name =
          ( class-in "_" size-name "_" short-dim "x" long-dim "in" ) |
          ( class-mm "_" size-name "_" short-dim "x" long-dim "mm" )

       class-in = "na" | "asme" | "oe" | "custom"
       class-mm = "iso" | "jis" | "jpn" | "prc" | "roc" | "om" | "custom"

    with a later section in the standard saying that additional class-in and
    class-mm names will be registered with the following syntax:

       class-in-name = ( lowalpha | digit ) *( lowalpha | digit | "." )
       class-mm-name = ( lowalpha | digit ) *( lowalpha | digit | "." )

    (b) The same questions are repeated for the first ABNF that appears in the
    PWG server (and which also appears in the standard as the first snap shot).

    The answers to (a) and (b) don't have to be the same. However, it may be
    less confusing if the ABNF can be the same for:
      (1) the ABNF for conformance to the standard
      (2) the first ABNF to appear on the PWG server, and
      (3) the snap shot copy of that first ABNF to appear in the standard
    Then this ABNF can appear only once in the standard.

    Tom

    -----Original Message-----
    From: Bergman, Ron [mailto:Ron.Bergman@Hitachi-hkis.com]
    Sent: Tuesday, July 10, 2001 17:20
    To: 'Hastings, Tom N'; Bergman, Ron
    Cc: ipp (E-mail)
    Subject: RE: IPP> MED - Proposed final ABNF for Media Size Self
    Describing Names

    Tom,

    But they will be in "a single production", since the ABNF will always be
    updated with the new registered values. I call that everyone agreed that
    the ABNF did not have to explicitly indicate that new values can be added.

    I like to hear from some developers that will "implement according to the
    ABNF" on this subject.

            Ron

    -----Original Message-----
    From: Hastings, Tom N [mailto:hastings@cp10.es.xerox.com]
    Sent: Tuesday, July 10, 2001 5:07 PM
    To: Bergman, Ron
    Cc: ipp (E-mail)
    Subject: RE: IPP> MED - Proposed final ABNF for Media Size Self
    Describing Names

    Ron wrote:

    I see you added "reg-class-xx-name" to the class-name definition. It seems
    like this definition is better handled as in D0.9. Otherwise, it could be
    interpreted to mean that arbitrary values can be added without registration.
    This is probably getting a little picky, but we want the ABNF to be a
    stand-alone specification.

    But if you remove the "reg-class-in-name" and "reg-class-mm-name"
    productions from the ABNF, then an implementation that follows the ABNF
    strictly, cannot accept a registered name that was registered AFTER the code
    was shipped. I think there will be objections from those who commented that
    all of the ABNF for conforming names should be in a single production, not
    in several pieces in the standard.

    Tom

    -----Original Message-----
    From: Bergman, Ron [mailto:Ron.Bergman@Hitachi-hkis.com]
    Sent: Tuesday, July 10, 2001 16:22
    To: 'Hastings, Tom N'; Bergman, Ron
    Cc: ipp (E-mail)
    Subject: RE: IPP> MED - Proposed final ABNF for Media Size Self
    Describing Names

    Tom,

    I spoke with Norbert yesterday concerning the syntax change to "custom" and
    he thought it was a good change. Since he probably has the most complete
    implementation using the media size names, I suspect this will not be an
    issue with anyone else.

    I agree that the ABNF should be included in the specification. But it
    should be noted that it is only a snap-shot in time and should never be used
    for more than an example. The "real ABNF must ALWAYS be obtained from the
    referenced file. I am not sure what you mean by...

    "Another reason to include the ABNF in the standard is that fetching the
    current ABNF from the web site isn't required (for an implementation or an
    implementer)."

    but it seems that an implementer would always want to see the latest. In
    practice, I don't expect to see many updates to the ABNF, but it should be
    pointed out that an update is possible. (I will add text that provides the
    appropriate warning.)

    I see you added "reg-class-xx-name" to the class-name definition. It seems
    like this definition is better handled as in D0.9. Otherwise, it could be
    interpreted to mean that arbitrary values can be added without registration.
    This is probably getting a little picky, but we want the ABNF to be a
    stand-alone specification.

            Ron

    -----Original Message-----
    From: Hastings, Tom N [mailto:hastings@cp10.es.xerox.com]
    Sent: Monday, July 09, 2001 2:46 PM
    To: Bergman, Ron
    Cc: ipp (E-mail)
    Subject: RE: IPP> MED - Proposed final ABNF for Media Size Self
    Describing Names

    I should have pointed out that the name field will now be required for a
    custom size name, as well as standardized names, whereas in Draft D0.9 the
    name field was optional in custom names.

    There are two advantages to requiring a name field in a custom size name:

    1. The syntax for custom and standard names becomes the same, simplifying
    implementation (since both will require a name).

    2. It is good practice for an administrator when defining a custom size to
    invent some sort of a name to suggest the size's use to its users.

    Ron,

    About your second question, I see I wasn't clear about whether or not the
    ABNF would appear in the standard or only in the PWG FTP site. I think that
    the first ABNF (current with version 1.0 of the standard) should be in the
    standard, including the commenting, so that implementations can see what to
    expect in case they choose to fetch the current (updated) ABNF from the PWG
    site.

    When we post the final approved standard, we should also post the ABNF flat
    file which will be identical to what is in the standard. Only when we
    register a new size name that needs a new class name, will we need to update
    the ABNF flat file on the PWG server. Periodically, but not more frequently
    than, say, once a year, we will update the standard with new size names and
    any new class names that have been registered in the meantime.

    Another reason to include the ABNF in the standard is that fetching the
    current ABNF from the web site isn't required (for an implementation or an
    implementer).

    Tom

    -----Original Message-----
    From: Bergman, Ron [mailto:Ron.Bergman@Hitachi-hkis.com]
    Sent: Monday, July 09, 2001 11:23
    To: 'Hastings, Tom N'; ipp (E-mail)
    Subject: RE: IPP> MED - Proposed final ABNF for Media Size Self
    Describing Names

    Tom,

    It should be pointed out that this does change the current syntax for custom
    size names, in that the "size-name" will be mandatory. This part is
    optional in the current draft.

    Also, I would like a clarification. Does this proposal completely remove
    the size ABNF from the document? Or, is the ABNF included with a note
    indicating where to get the latest? Although it may be somewhat confusing,
    it would be cleaner to only include the reference to the ABNF.

            Ron

    -----Original Message-----
    From: Hastings, Tom N [mailto:hastings@cp10.es.xerox.com]
    Sent: Monday, July 09, 2001 10:57 AM
    To: ipp (E-mail)
    Subject: IPP> MED - Proposed final ABNF for Media Size Self Describing
    Names

    Ron and I want to publish a final version of the Media Standardized Name
    standard for Last Call. There are three standards that require its use, so
    we need to complete the Media standard. One is UPnP Basic Printing
    Template. Another is IPP FAX.

    Please send any comments on this proposed ABNF BEFORE we publish the next
    draft. Silence will be interpreted as agreeing with the ABNF.

    ****************************************************************
    Please send any comments by Tuesday, July 17, 2001.
    ****************************************************************

    There has been the suggestion on the mailing list that the ABNF should be a
    single production, including the custom part and the future registered class
    name part as is common practice.

    We are also agreeing to retain the explicit list of class names in the ABNF.
    When a new size name is registered which needs a new class name, then the
    ABNF will be updated to include the new class name. The updated ABNF will
    be available on the PWG site as a flat text file in a published and stable
    URL.

    Advantages of this approach:

    1. All of the ABNF for the Media Size Self Describing Name is specified as a
    single ABNF production, so that code can be written that will be stable,
    even as new media size names are registered with new class names. This is
    the common practice for other standards that use ABNF.

    2. Which class names go with which units is part of the ABNF, so that it is
    clear that using the wrong set of units with a class name violates the ABNF
    and permits a parser to detect such mal-formed ABNF.

    3. Using a single "custom" class name (rather than one for in and one for
    mm) simplifies the parsing for custom names because the parser needs to look
    for one special prefix ("custom"), as with Media Type and Media Color,
    rather than for, say, "custom-in" and "custom-mm" class names. Since
    administrators are defining the custom sizes, not vendors, whether they
    choose in or mm units will depend on their users.

    Disadvantages of this approach:

    1. When new class names are registered, the existing parsers won't be able
    to validate that the units are correct, unless they access the PWG web site
    for the current ABNF. (The standard will NOT require implementations to
    access the PWG web site to get the current ABNF, nor does it require that
    implementations detect corrupt names that use the wrong units).

    The ABNF flat file will have comments at the beginning to identify the date
    and version of the standard and the date and version of the ABNF. The ABNF
    date and minor version number will be updated every time it is updated. The
    ABNF follows RFC 2234.

    Here is the proposed final ABNF as it will appear in the flat file on the
    PWG site in:
    ftp://ftp.pwg.org/pub/pwg/standards/pwg5101-abnf.txt:

    ; ABNF for the Media Size Self-Describing Name
    ; Part of the IEEE/ISTO 5101.1 Media Standardized Names standard
    ; For use with the IEEE/ISTO 5101.1 Standard Version 1.0, July 9, 2001
    ; ABNF version 1.0, July 9, 2001
    ; This ABNF is updated whenever a new media size class name is registered
    ; according to the procedures of IEEE/ISTO 5101.1.

       media-size-self-describing-name =
          ( class-in "_" size-name "_" short-dim "x" long-dim "in" ) |
          ( class-mm "_" size-name "_" short-dim "x" long-dim "mm" )

       class-in = "na" | "asme" | "oe" | "custom" | reg-class-in-name
       class-mm = "iso" | "jis" | "jpn" | "prc" | "roc" | "om" | "custom" |
                  reg-class-mm-name

       reg-class-in-name = ( lowalpha | digit ) *( lowalpha | digit | "." )
       reg-class-mm-name = ( lowalpha | digit ) *( lowalpha | digit | "." )

       size-name = ( lowalpha | digit ) *( lowalpha | digit | "-" )

       short-dim = dim

       long-dim = dim

       dim = integer-part [fraction-part] | "0" fraction-part

       integer-part = non-zero-digit *digit

       fraction-part = "." *digit non-zero-digit

       lowalpha = "a" | "b" | "c" | "d" | "e" | "f" | "g" | "h" | "i" |
                  "j" | "k" | "l" | "m" | "n" | "o" | "p" | "q" | "r" |
                  "s" | "t" | "u" | "v" | "w" | "x" | "y" | "z"

       non-zero-digit = "1" | "2" | "3" | "4" | "5" | "6" | "7" | "8" | "9"

       digit = "0" | "1" | "2" | "3" | "4" | "5" | "6" | "7" | "8" | "9"



    This archive was generated by hypermail 2b29 : Wed Jul 11 2001 - 18:38:33 EDT