While I'm not entirely unsympathetic to Martin's position that all RTL
scripts should be grouped together, I think we're getting off the topic
slightly. I'd like to make a couple of points.
1) The directionality assigned characters should *not* be done in a
fashion to be convenient to programmers. It should be assigned on the
basis of the actual way the script itself works or worked.
I should point out, BTW, that we *already* have broken the assumption
that all RTL characters are in a contiguous block -- the Hebrew wide
letters in the FBxx range are RTL, as are the Arabic compatibility forms.
2) The impression I got from the Etruscanist who replied to Michael was
that modern Etruscan materials are written LTR as a result of the
underlying technology, not out of a genuine preference on the part of
Etruscanists. (Michael, do you think that's a fair assessment?) Most of
the Etruscan material I've been able to locate has been LTR but also in
Latin transcription.
If my gut feeling is right, then to encode Etruscan LTR on this basis
would be rather like encoding everything as precomposed because the
technology most people are currently using has an easier time of that. I
don't think such considerations should be ignored, but I don't think they
should be paramount, either.
In this case, the relevant question to ask is if an Etruscan scholar
preparing a modern summary of Etruscan texts *in Etruscan* would use the
direction of the original text or not. If so, then RTL is probably the
better way to go. (Michael, do you know what the _Corpus of Etruscan
Inscriptions_ does?)
3) As we encode more and more ancient scripts, we're going to find more
and more challenges to our assumptions that characters have well-defined
properties and well-defined semantics. Directionality is a good case in
point. Historically, Etruscan was written boustrophedon, then RTL, then
LTR -- but even during the period when it was predominantly RTL, it would
be written LTR if it were more convenient to do so.
The current Unicode architecture allows for this to be properly handled.
We define a default direction and allow directionality overrides to be
used. **The main thing is to have a rendering rule to the effect that
Etruscan characters flip their shapes when their direction changes.**
My own personal prejudice is towards encoding Etruscan as RTL because
that's kind of the "majority use" of the script in antiquity, but not by
a terribly wide margin. Methinks that if one were to actually ask an
ancient Etruscan which direction was best, they'd shrug their shoulders
and get on with life.
=====
John H. Jenkins
jenkins@apple.com
tseng@blueneptune.com
http://www.blueneptune.com/~tseng
This archive was generated by hypermail 2.1.2 : Tue Jul 10 2001 - 17:20:37 EDT