RE: Re[2]: marks (2 new symbols)

From: Philippe Verdy (
Date: Tue Oct 02 2007 - 11:22:32 CST

  • Next message: Mark Davis: "Re: Proposal for additional syntax (was Re: New Public Review Issue: Proposed Update UTS #18)"

    Dmitry Turin wrote:
    > Computer in both field of rational database and full-text database
    > must compare only 'sense' itself of text.
    > Let me refer once again to rough equality of strings

    Please stop referencing your *own* external references, always on your own
    website (which contains lots of errors, and a very confusive and
    inconsistent terminology, the inconsistence being the worst problem, because
    a confusive terminology would not be a severe problem, provided that you use
    it consistently and with precise definitions) as if they were proofs of

    Really you need to make progresses in English, or you should work in Russian
    to make sure that your proposals are consistent and it's difficult for you
    to use a non-confusive English terminology.

    Why don't you start by reading the Unicode standard to get a stable
    terminology with their definitions?

    I completely disagree your opinion that a "full-text database *must* (sic!)
    compare one 'sense' itself (sic! Do you mean semantic?) of text. There's no
    such requirement in databases: a database does not compare anything, you're
    mixing it with a database application, which is not even required (your
    "must") to compare things semantically:

    In most database applications this is completely false, as entities are
    compared in most cases for strict equality, to implement relations between
    equal keys in the data model (and in database models with normalized forms,
    theses keys are not even encoded as text, but as numeric entities or
    symbolically as strict identifiers, with a complete absence of semantic
    except being a unique identifier, whose internal encoding does not matter as
    this is transparent data).

    Weak relations based on semantic relations are NOT implemented in almost all
    relational database engines. What is really implemented is named "collation"
    and it is not used for building the data model but for performing queries
    with "range" constraints.

    This archive was generated by hypermail 2.1.5 : Tue Oct 02 2007 - 11:25:45 CST