W3C home > Mailing lists > Public > public-rif-wg@w3.org > May 2008

text on XML syntax extensibility (ACTION-470)

From: Sandro Hawke <sandro@w3.org>
Date: Mon, 19 May 2008 18:19:52 -0400
To: public-rif-wg@w3.org
Message-ID: <14455.1211235592@ubuhebe>

Here's some text for BLD which says what I think we need to say about
syntactic conformance for extensibility.   

    Systems which accept RIF input in the Basic Logic Dialect are
    required to check whether the input conforms to the grammar of BLD
    and (unless overriden by the user) to reject input documents which
    fail the check.  This check can be done by an XML Schema validating
    parser or by other software.  This check is required as part of the
    RIF approach to multiple dialects, extensions, and fallback
    translations.  If, instead of rejecting the document, a BLD system
    "repaired" it (such as by ignoring the part it did not recognize),
    it could silently produce incorrect results if it were (quite
    reasonably) given a RIF document which used some other dialect or
    some extension.  Given the undesirability of silent failures under
    normal use, systems which accept BLD as input MUST reject any RIF
    document which is not syntactically valid BLD.

    The rejection message SHOULD inform the user that the input does not
    conform to the syntax of the Basic Logic Dialect and that it may be
    a valid RIF document in some other dialect or using some extension.
    The system MAY allow the user to override the rejection and proceed
    with some ad hoc repair process, with clear warnings about the
    possibility of incorrect results.  The system MAY inform the user
    that it was unable to perform any fallback translation to BLD, but
    this messaging is not required in this version of BLD.

    EDITOR'S NOTE: The preceding text is expected to change once the
    fallback translation system is specified.  At that point, a new
    requirement is expected to be put into place, that systems SHOULD
    implement the fallback translation system, and that if they do not,
    they MUST inform users that they do not implement this RIF feature.

Does that make sense?     This completes ACTION-470 (give or take
whatever discussion we need to have on this).

    -- Sandro
Received on Monday, 19 May 2008 22:22:28 UTC

This archive was generated by hypermail 2.4.0 : Friday, 17 January 2020 17:07:44 UTC