review 1

From: Aaron Chang (anc327@yahoo.com)
Date: Sun Apr 18 2004 - 17:04:33 PDT

  • Next message: Joe Xavier: "Review : Relational Databases for Querying XML documents ..."

    The Essence of XML
    by Simeon and Wadler

    Review Points
    =============

    1. The authors of this paper seek to study two major
    XML grammar issues:

            A. relation of named vs structural types
            B. relation of matching vs validation of data

    2. external formats for representing data require
    self-description and round-tripping,
       neither of which are in XML.

    3. authors seek to clarify "the essence of XML" being
    that of data typing and validation

    4. XML has many typing systems. the authors suggest a
    formalization of XML Schema typing system.
       Schema uses XQuery and Xpath semantics.

    5. XML does not match a value to a type, it validates
    (it happens or it fails). authors present
       a theorem that validation occurs IFF a value
    matches a type and "erases" to an untyped value.

       this simple statement appears to be an important
    distillation of fact, learned by trial/error

    6. Schema is written in XML notation. the authors
    place specific restrictions on the notation
       to make it more readible. the simplifications
    offered sound reasonable.

    7. some important aspects of their model on Schema:
            A. simple vs complex types
            B. global vs local declarations
            C. atomic, list, and union types
            D. "derivation" of simple and complex types (like
    subclassing) -> called "restriction"
            E. derivation by extension

            conclusion: should restriction and extension be
    essentially the same grammar?

    8. specification of different types and untypes and
    their values is well-thought out

    9. "yields" and "resolves" are good, but not very
    intuitive, perhaps "instanceOf" or "isType" like
       in Java would've been better

    10. not sure why "substitutes" is necessary, why not
    just use "=" ?

    11. what's the difference between "derives" and
    "resolves"?

    12. "matches" looks useful, but perhaps "isType"
    could've also worked not only between types but also
         between values and types

    13. erasure and validation are important contributions
    by the authors

    14. the validation formal theory is very useful and
    intuitively presented

    15. sensibility in some cases seemed redundant, esp
    subtyping

    overall a very interesting read on how to place a
    grammar on a largely unregulated semantics

            
                    
    __________________________________
    Do you Yahoo!?
    Yahoo! Photos: High-quality 4x6 digital prints for 25¢
    http://photos.yahoo.com/ph/print_splash


  • Next message: Joe Xavier: "Review : Relational Databases for Querying XML documents ..."

    This archive was generated by hypermail 2.1.6 : Sun Apr 18 2004 - 17:04:41 PDT