RE: Version(s) of Unicode supported by various versions of Microsoft Windows

From: Kenneth Whistler (kenw@sybase.com)
Date: Fri Mar 05 2004 - 19:28:01 EST

  • Next message: Peter Constable: "RE: A proposed change of name for Latin Small Letter THwithStrikethrough"

    The Peters continued:

    > > What, not even a product whose only functions are to convert between
    > > Unicode encoding forms?
    >
    > OK; I guess I misunderstood the question you were asking. I see now that
    > it's "Can a given product correctly perform all of its functions
    > (whatever those might be) for all characters in a given version of
    > Unicode?" Sure, there will be combinations of products and versions of
    > Unicode for which that will be true. Character Map is probably one such
    > example.
    >
    > But if the product is an OS, I'm pretty sure there will not be any.

    And I think this illustrates one of the difficulties that plagues
    such questions in the software industry today.

    What is "a product"? And what is "a system"?

    These notions used to be much simpler when computers used to all
    be discrete, OS's were much smaller and more primitive, and
    software applications were either small and single-function,
    or were massive, custom-developed applications for mainframes.

    Nowadays, everything is massive, multi-tiered, distributed,
    and interdependent. "Product" definitions are often constructed
    by marketing teams playing component legos, rather than being
    based on some kind of standalone executable piece of software.
    In these contexts, what constitutes "Unicode support" can get
    lost in the forest.

    Conceptually, it is a little like asking the question, "Does
    the Internet support Unicode, Version X.X?"

    As Peter [C] has indicated, if you properly narrow down your
    concept of what a "product" is and what "support" for Unicode it has,
    in some meaningful subset of what support might mean, then you
    can get to a well-formed question that might have an unambiguous
    yes/no answer. Whether an *answerable* question of this sort
    is then actually interesting for the person who was inquiring
    in the first place, is itself another question.

    For example, I develop a Unicode library that I am *positive*
    has full Unicode support for all characters for all versions
    of Unicode (and all future versions) for its basic string
    copy and length operations. But I doubt that answer will be
    of much real interest.

    When you move on to more meaningful questions such as does it
    support default casing for all Unicode characters, then I have
    to start hedging the answer carefully with which particular
    version of Unicode, which particular version of the library,
    and which date. And applications which build on the library
    have such dependencies built in, without it necessarily being
    obvious to all the developers using the library and certainly
    not to the end users of the distributed software that might
    be visible to them.

    --Ken



    This archive was generated by hypermail 2.1.5 : Fri Mar 05 2004 - 19:54:27 EST