From: Asmus Freytag (asmusf@ix.netcom.com)
Date: Fri Dec 27 2002 - 03:01:16 EST
At 10:09 PM 12/25/02 +0330, Roozbeh Pournader wrote:
> > In fact the glyph for Kurdish Q often looks like a large q, similarly to
> > Cyrillic h; this is an inappropriate glyph for Latin Q.
>
>This should be enough evidence. Any samples?
OK (assuming that this can be substantiated by samples) we now have the
following score
Legacy data.
Legacy data would tend to have been created with Latin/Cyrillic character
sets, which unify the Latin and Kurdish q. Disunification would make
conversion *from* legacy data context dependent. (Conversion to legacy sets
could be done by mapping both types of Q to the same value, i.e. Latin Q,
in the legacy set). However, the prospect of forever seeing both Latin and
Cyrillic Q used in Kurdish text, due to contamination from legacy data
leads me to award on negative point to the disunification proposal:
Score -1
Fonts
If the 'q' form of capital Q is any common, maintaining the unification
would introduce a language dependency in the rendering process. If the 'Q'
form is not wholly un-acceptable, then its arguably a permissible
fall-back. Standard fonts lacking language specific glyphs would be usable
(after a fashion) for Kurdish, while the reverse is not true. However, the
font technology to handle a language specific form of capital q clearly
does exist. However, Latin/Greek/Cyrillic rendering normally does not
require such mechanisms for rendering 'plain' text.
Score: 0.5
Sorting
Having the letter q both be part of Kurdish words (in Cyrillic) and Latin
words, would make sorting of mixed lists a bit tricky. I'm not sure how far
the level of required trickery exceeds some of the trickery one needs to
apply for unusual cases in other languages (e.g. the case in Danish where
'aa' can either be 'aa' (if a across an intra-compund word boundary) or
a-ring).
However, on the priniciple that disunification would simplify the sorting
problem I'd award a 1/2 point.
Score: 0.5
In summary, with the information on capital q the score tends to even out.
If evidence could be adduced that a) no legacy data exists, or b) the user
community wishes this problem on itself, a proposal could acquire a small
positive score, possibly enough to overcome inertia.
A./
This archive was generated by hypermail 2.1.5 : Fri Dec 27 2002 - 03:30:14 EST