From: Asmus Freytag (
Date: Fri Oct 21 2005 - 21:41:15 CST

  • Next message: Jukka K. Korpela: "Re: LAO LETTER FO SUNG and LAO LETTER FO TAM"

    On 10/20/2005 8:58 AM, Edward H. Trager wrote:

    >The decision to make Unicode Normative Names immutable is similar
    >to doing one of the following:
    > -- Printing Webster's Dictionary of the English Language but having the
    > publisher repeatedly refuse to fix errata in subsequent printings.
    > -- Publishing a significant new scientific finding in the journal Nature,
    > but then having the journal's editor refuse to publish a subsequent notice of
    > errata or notice of retraction.
    > -- Writing the Constitution of the United States, but not permitting
    > the 13th Amendment (Abolition of slavery) or any other amendments
    > crucial to the realization of a true democracy in the American nation.
    >Would anyone even buy Webster's if the publisher did that?
    >Would any scientist even bother to read Nature if the journal did that?
    >Am I opening up a can of worms?
    Your comparisons are not as apt as you imagine.

    There is always a tension between formal identifiers and human use of
    them as mnemonic labels. Would the C-runtime be more or less successful,
    if each vendor had vetted the names of the functions and renamed those
    that were misleading?

    How about the order of parameters supplied to these functions - they are
    downright inconsistent, and do lead to programming errors -- yet no-one
    wants to re-architect that API.

    All these are environments where absolute stability allows existing
    software to continue to operate, even if the burden of navigating the
    inconsistencies is placed on the human user.

    The Unicode character names were designed for use as (shared)
    identifiers across a series of ISO standards - the same name is used in
    several standards for the same character, even if the character is at
    different positions in each. The use of these identifiers outside this
    ISO environment may not have materialized to the degree its designers
    (predating Unicode) had hoped.

    They certainly could not have envisioned the degree to which users would
    try to rely on these labels as a source of information *about* the
    character in question. However, that doesn't make their design invalid,
    nor does it make the later decision (supported by Unicode) to take the
    logical conclusion from the original design parameters and make the
    character names formally immutable. Not doing so would have left only
    the code position as a unique identifier, and that lacks the necessary
    redundancy for such a large set f character.

    One side effect of this stability is that it is now possible to document
    issues about characters in a human readable way that is stable. You can
    now write: "LATIN CAPITAL LETTER AE is really a ligature" and not have
    to update this sometime in the future when the committee might decide
    suddenly that the name should reflect this rather than that aspect of a

    The other side effect is that it has enabled the UTC to get on with
    business, since the number of requests to change character names based
    on preference would have been a serious drain on resources -- instead,
    they either result in editorial addition of a comment, or a footnote in
    the meeting minutes.

    >The decision by the Unicode Consortium to *not* provide normative names for
    >the Unified Han characters represents a precedent of either *inconsistency*
    >or *amendation* in the application of the rule requiring
    >assignment of normative names in Unicode.
    The normative name for character 4E00 is CJK UNIFIED IDEOGRAPH-4E00 and
    so on. Your statement is contra-factual. Given that the names are
    predictable from the code position (as are the names for Hangul
    syllables, even though they do not contain the code point in the name),
    it would be a waste of resources (paper, and the money of people buying
    the Unicode book) to have printed a full listing of all 80,000 of such

    The rest of your argument, being built on top of a false premise,
    collapses entirely, therefore I'm not going to respond to any particular
    item in it.


    PS: one more comment below.

    >It seems that the Consortium has
    >said, "yes, on the one hand normative names must be assigned, and once
    >assigned they are immutable. But on the other hand, there are just too many
    >Han characters, so let's *amend the rule* for the special case of Han characters
    >so that we won't have to assign individual normative names to them."
    >Is the argument sound?: A precedent appears to already exist where
    >the rule of assigning immutable normative character names was amended so that
    >normative character names are not assigned for Han characters.
    >Therefore, a new *amendment to the rule* could be added to allow for corrections
    >of normative names in cases of obvious clerical or editorial mistake, ... as long as
    >approved by two-thirds majority et cetera et cetera ...
    >The designers of the American constitution realized that amendments would be
    >necessary to the well-being of a nascent democracy. Is it not possible to
    >persuade the Unicode Consortium and ISO/IEC 10646 that amending the rule of normative
    >name immutability in cases of obvious error is simply in *everyone's long-term
    >best interest*, and not doing so is either sheer short-sightedness or plain stupidity
    >which may result in the continuation of lengthy and annoying debate ad infinatum that
    >hinders the otherwise excellent and important work of this standard?
    >In conclusion, let me note that it is especially evident in this age of electronic
    >communication that there exist very successful models of cooperative endeavour
    >in which the distinctly human problem of error and the correction of that error is
    >handled efficiently without the fuss and ennui of endless committee meetings, or
    >even endless postings on mailing lists. For example, there is Wikipedia.
    >( Wikipedia is wide-open. It is completely *promiscuous*. So why is it growing so fast,
    >and why is the quality of the articles consistently so high? Why isn't it just
    >full of complete nonsense? : )
    The wikipedia is not a standard.

    >Is it possible for all of the brilliant men and women who have brought us the Unicode Standard
    >to enter *deep thought* and consider the following : Are there ways to mix the traditional
    >committee-based decision-making systems used by the Unicode Consortium, ISO/IEC 10646, and
    >standards organizations the world over with
    >some of the novel but clearly effective methods of systems like Wikipedia ?
    >- Edward Trager

    This archive was generated by hypermail 2.1.5 : Fri Oct 21 2005 - 21:42:15 CST