Re: unicode in emacs 21

[Dave Love]

>>>>> "rms" == Richard Stallman <rms@gnu.org> writes:

 rms> I think that supporting Unicode at the internal level is the
 rms> best way to support it fully, and that's what we have decided to
 rms> do.

How is handa's scheme for simply adding 20-bit Unicode to the current
internal encoding at odds with that?

 rms> As a result of that decision, we are sometimes reluctant to put
 rms> time into studying, installing and maintaining other approaches
 rms> which would be obsolete once we do it the right way.

I've seen no convincing rationale for this `right way' versus an easy,
backward-compatible extension.  It might even be forward-compatible.

 rms> Supporting Unicode superficially 

I'm not sure I'm suggesting anything superficial, just adding the
necessary Unicodes, which is what is currently missing.

If there were enough private charsets free now, you could probably
have a pretty good Unicode editor already, modulo bidi.  (From the
point of view of actual users.)

 rms> while retaining the current internal representation raises a
 rms> number of problems, one of them being that the internal
 rms> representation has several alternatives for the same character
 rms> which correspond to the same code point in Unicode.

Having been working with it, I have no idea why such redundancy is a
problem in practice; one can canonicalize as necessary.  Your
preferred design apparently also has it anyhow, apart from unicode's
multiple representations of the same character (see `normalization').