Note: I don't advocate "duplicate encoding" as you think. But probably the
current IDS model is not sufficient to describe characters correctly, and
that it may be augmented a bit (using variant codes or some additional
joiners or diacritics?).
But IDS strings are suitable for rendering as ligatures and this should be
permitted, and should even be the standard way to represent personal names
without making them depend on an unproved single distinctive presentation.
E.g. someone writes his name with some personal strokes and uses it as its
registered "signature"; he is then doing business or is cited in news with
simplified presentation, and the Chinese authorities also use their own
simplications. All these will designate the same person. But who is correct
for the presentation of the character ? In my opinion it is only the person
that invented it for themselve, as a personal signature, but this is not
suitable for encoding (privacy and copyright issue). All the other
presentation are legitimate, and we don't need additional encoding for it:
the ligaturing of IDS strings is sufficient even if it does not match
exactly the person's signature.
2018-03-07 23:04 GMT+01:00 Philippe Verdy <verd...@wanadoo.fr>:
> I'm just speaking about the many yearly inventions of sinograms for
> personal/proper names, not about the ues of traditional characters for
> normal language.
> People just start by assembling components with common rules. Then they
> enhance the produced character just like we personalize signatures. But for
> me, all these look like personal signatures and are not neede for formal
> encoding and even these persons will accept alternate presentations if it's
> just to cite them (and would not like much that you imitate their personal
> signature by standardizing it in a worldwide standard: I think many of
> these encodings have severe privacy issues, possibly as well copyright
> issues !).
> 2018-03-07 22:35 GMT+01:00 Ken Whistler <kenwhist...@att.net>:
>> On 3/7/2018 1:12 PM, Philippe Verdy via Unicode wrote:
>>> Shouldn't we create a variant of IDS, using combining joiners between
>>> Han base glyphs (then possibly augmented by variant selectors if there are
>>> significant differences on the simplification of rendered strokes for each
>>> component) ? What is really limiting us to do that ?
>> Ummm.... ambiguity, lack of precision, complexity of model, pushback by
>> stakeholders, likely failure of uptake by most implementers, duplication of
>> representation, ...
>> Do you think combining models of Han weren't already thought of years
>> ago? They predated the original encoding of unified CJK in Unicode in 1992.
>> They weren't viable then, and they aren't viable now, either, after 26
>> years of Unicode implementation of unified CJK as atomic ideographs.