RE: [idn] stringprep comment 5: hangul conjoining sequence

["Kent Karlsson" <kentk@md.chalmers.se>]

> -----Original Message-----
> From: Soobok Lee [mailto:lsb@postel.co.kr]

> ----- Original Message ----- 
> From: "Kent Karlsson" <kentk@md.chalmers.se>
>  
> > >  3. Compatibility hangul jamos are mapped into conjoining 
> > > jamos without any fillers.
> > 
> > Compatibility (non-conjoining) Hangul letters are best prohibited.  Doing
> > the correct mapping is not expressible in via nameprep without adding
> > a new, special for Hangul, mechanism.  Would there be any major problems
> > just prohibiting them? (Allowing only the conjoining jamo and syllable characters.)
> > 
> 
> That's impossible. Unicode-issued KSC5601 ==> UNicode mapping tables
>  provides "hangul legacy code jamos --> Unicode u+3??? 
> compability jamos".

Yes. 

>  When ordinary Korean users enter hangul jamo sequence in 
> all version of Windows or LInux which adop the table above, 
> those legacy-coded jamo sequences will be eventually converted into 
> Unicode compatibility jamos sequence. There is no way to 
> input directly  conjoining jamo sequences
> including necessary fillers and that will be painfully 
> complex to end users.

Keyboard drivers+IME handling for Hangul often take Hangul compatibility
letters and (when the IME deems appropriate) **replaces** subsequences
of (maybe a Hangul syllable character and) Hangul compatiblity letters with
a Hangul syllable character.

It is true that many IMEs may not be able to produce all and any Hangul syllable,
in the logical sense, not even all and any Hangul syllable Johab character.  But
that is beside the point here.

> And Even windows and linux does not render correctly the 
> conjoining jamo sequence including fillers.

True, but that is a lack of support issue with those systems, and should be
taken up in some other forum.

> The compability jamos are the only viable alternative for 
> correct jamo sequence rendering in
> commerical windows and linux platforms.

The compatibility letters render as spacing characters; they do not normally
conjoin into syllable blocks.  Doing that analysis (with a filler to tell that a
"compatibility 'maybe conjoning'" sequence begins) of the input and proper
conversion to Hangul conjoining jamo is still beyond what nameprep can do
with just simple additions to the tables. Adding a proper, and Hangul specific,
conversion step *to nameprep* seems inappropriate.  This is an input issue.
But yes, that conversion may be useful, but I would not push it into nameprep.

A "better keyboard" could generate the conjoining single-letter jamos directy,
without using any IME (17 consonants (twice, via shift: lead, trail) and 11 vowels
plus some very few historic variants fit easily on a keyboard).  Indeed, there would
then be no problem in typing all and any Hangul syllable (in the logical sense), as
long as the result could be fitted into a rendered syllable block. But such a keyboard
would be appropriate only after the display issue became obsolescent; which it
isn't yet.  However, it may someday relatively soon be an obsolescent rendering
problem.  (Netscape 6 is close to rendering them correctly already:
http://jshin.net/i18n/jamo.html)

		Kind regards
		/kent k


> Soobok 
>