L2/09-256

-- -- -- -- -- -- -- -- -- -- -- -- -- -- -- --

Date/Time:    Sun Jul 26 05:32:15 CDT 2009
Contact:      mattias.ellert@fysast.uu.se
Name:         Mattias Ellert
Report Type:  Feedback on an Encoding Proposal
Opt Subject:  Comments on ISO/IEC JTC1/SC2/WG2 N3555 / L2/08-428

Hi!

Comments on ISO/IEC JTC1/SC2/WG2 N3555 / L2/08-428

The names of the following characters in the proposal are inconsistent
with the names of similar characters already encoded:

   1E0AE   LATIN SMALL LETTER NASAL Y
   1E0B2   LATIN SMALL LETTER NASAL SCRIPT Y
   1E0B4   LATIN SMALL LETTER NASAL OPEN Y

They should be changed to follow the same pattern as other characters
with an ogonek modifier:

   1E0AE   LATIN SMALL LETTER Y WITH OGONEK
   1E0B2   LATIN SMALL LETTER SCRIPT Y WITH OGONEK
   1E0B4   LATIN SMALL LETTER OPEN Y WITH OGONEK

Landsmålsalfabetet uses an ogonek to indicate nasal vowels. This is done
for any character representing a vowel, including the different y's.
Due to the desender on of the y's it is tricky to find an anchor point
and the ogonek is usually slightly tilted when attached to a y in order
to attach properly when compared to other letters with an ogonek.
However, logically this tilted ogonek is still an ogonek, and should
be called and encoded as such.

These letters should also have a canonical decomposition to the base
letter + a combining ogonek, like all other letters with ogoneks in
the standard. It is very likely users will enter these sequences to type
these letters since this is what they will do for all other nasal letters.
If there is no such decompositions there is no unique spelling for
these characters.

   1E0AE =  0079 + 0328  # LATIN SMALL LETTER Y + COMBINING OGONEK
   1E0B2 = 1E0AF + 0328  # LATIN SMALL LETTER SCRIPT Y + COMBINING OGONEK
   1E0B4 = 1E0B3 + 0328  # LATIN SMALL LETTER OPEN Y + COMBINING OGONEK

Since the base letter for the first combination is already in the standard
that combination would have to be added to the composition exclusions in
order to preserve normalisation stability.

Current practise is to not add new precomposed letters that can be represented
by a combining sequence, unless there is a really good case for it. The only
case that could be made for adding these letters is to make font designers
aware of the need to slightly tilt the ogonek when combined with these
glyphs, but that argument is admittedly weak. Hence, the correct course
of action might be to drop the three precomposed letters from the proposal
and instead mention the need for tilting the ogonek when combining it with
base letters with a desender somewhere else in the standard.

Various legacy fonts used for the landsmålsalfabetet usually encode a
combining ogonek that is used to mark nasal vowels. In addition - since
these fonts usually do not use modern context dependent glyph shaping -
they also provide precomposed glyphs for the different y's with ogonek,
since the combining ogonek does not attach very well in these cases without
context dependent shaping. A modern font with context dependent glyph shapes
can tilt the ogonek automatically, and no special code point should be needed
for the tilted ogonek or precomposed letters that use it - provided the font
designer has done of good job.

-- -- -- -- -- -- -- -- -- -- -- -- -- -- -- --