Re: DIY OpenType Re-ordering

From: Antoine Leca (Antoine10646@leca-marti.org)
Date: Fri Mar 24 2006 - 12:03:41 CST

  • Next message: Jon Hanna: "Re: Case insensitive comparisons"

    Philippe Verdy wrote:
    > 919, 94D : dead "NG" (NGA + VIRAMA):

    More exactly, "dead NGA", or "NGAd". OK.
    So you agree this is one, identificated, combing character sequences (KI
    being another).

    > uses full NGA with subjoined halant by default,
                                          ^^^^^^^^^^
    Huh?

    > or half-form NGA if followed by consonnant,
         ^^^^^^^^^^^^^
    Beg your pardon, but I do not know such a thing, neither what could be a
    "half-consonant form of NGA," NGAh.

    > or half-form NGA, unless ligatured with next consonnant.

    Cannot make sense of this.

    > 915, 93F : live "KI" (KA + I): uses full-KA form, unless ligatured
    > with previous consonnant.

    Live KA, or KAl, which has the nominal form of KA, + Ivs; or Ln + Ivs. Yes.
    And here is precisely the point: the rules for Devanagari are based upon the
    ligatures (or not) between the various consonants in a cluster, which appear
    to group several combining character sequences. At such, introducing a rule
    speaking about c.c.s. to define reorderings does not seem perfect to me, it
    merely confuse things, IMHO.

    Or it requires to define a whole new concept, here "something which
    *applies* to a (or several) c.c.s."
    No definition for "to apply" found, back to square one.

    > The choice between full-form NGA with subjoined halant or half-form
    > depends on locale,

    Well, Mr. Karlsonn was asserting in a previous post that any possible
    variation here would depend on
         A *spelling* difference that should be recorded in the
         sequence of characters (in some, not yet standardised,
         way), quite apart from font issues.

    Sorry, but I see a very big difference between the two positions.

    > but by default NGA and KA do not ligate,
          ^^^^^^^^^^
    More huh!?
    Please send me the quote from your official book about Nagari (Sanskrit)
    Manual of Style saying so.

    Please do you point me toward Hindi material. Only Sanskrit.
    (I am hearing that Hindi will not use NG in this position, rather anuswar on
    the preceding consonant; but I cannot be completely affirmative here, so
    please double check this information, or give me authoritative statements;
    thanks in advance.)

    Also have a look at 9-2.

    > The vowel I (appearing on right) is reordered before consonnants of
    > the cluster,

    Again, this is precisely the point under question.
    I know this is what states TUS (but see below about this).
    However, the fact is, it does not completely match with the observations of
    the uses in India. So what? Is Unicode in the business to dictate to Indians
    how they should write their own script?

    [ Philippe knows quite well about a raging debate in France with similar
    roots and behaviours. ]

    > This is fully described in Unicode...

    What is fully clearly spelled out in Unicode is that rendering is not
    specified by the Standard!

    The paragraph you are probably referring to starts with (at the end of the
    first line, on page 224 in TUS4) the statement there are the rules for a
    *minimum* rendering of Devanagari.

    And this was the very point I was pointing out to Vinod in the first place.
    Sorry if my post is a bit acerb, but I am more or more feeling what I wrote
    is not even read, just commented...

    Furthermore, these rules can be improved, obviously. /This/ was part of
    Vinod's original point, and I agree about the need for improvement
    (particularly about the other scripts, but Nagari also should be looked at.)
    Also, about your detailled point, I do not see rule R15 be cancelled for
    Tamil (only script with details)... Again, I agree with Vinod, the level of
    the Standard is insufficient here. And now you try to explain me that it
    should be imposed to override Indians' habits...

    > If you insert a ZWJ in the middle [...]
    > Anyway, the vowel I is still reordered before the half-form NGA.

    How does it links with c.c.s.?
    Why is it different among Indic scripts?

    > If you insert a ZWNJ, it blocks the reordering of I before dead NGA,

    How does it links with c.c.s., and "to apply"?

    Antoine



    This archive was generated by hypermail 2.1.5 : Fri Mar 24 2006 - 12:05:09 CST