From: Kenneth Whistler (kenw@sybase.com)
Date: Fri May 21 2004 - 15:01:33 CDT
Peter Kirk suggested:
> Similarly, I suppose, with the proposed Phoenician script: each
> character could be given a compatibility decomposition to the equivalent
> Hebrew letter. This implies automatic interleaved collation. Now, while
> I don't expect Michael Everson to jump at this suggestion,
Not only will Michael Everson not jump *at* this suggestion, I
will jump *on* it.
We've already had the (tedious) discussion about interleaved
collation. Whatever the merits of interleaving two related
scripts in the Default Unicode Collation Element Table, at least
in that context the equivalencing of weights is entirely a
matter for the UCA (and ISO 14651) collation standards.
But proposing compatibility decompositions to do this automatically
would drag *normalization* into the fray as well. And for this
particular proposal you'd end up with two encoded scripts
which would fold together for some normalization forms (NFKC
and NFKD) and would *NOT* fold together for some normalization
forms (NFC and NFD). I would consider that very unintuitive
and bad behavior -- likely to introduce more problems than it
attempts to solve.
There have been no precedents in the standard where *normalization*
attempts to equate two scripts in the standard. That is considered
to be the province of transliteration, rather than normalization.
I think the chances that the UTC would seriously consider a
proposal to give compatibility decompositions to an entire
script equating it to another script to be very low indeed.
> I do think
> that it should be considered as a compromise between the otherwise
> irreconcilable desires of many to unify the two scripts and of some to
> separate them.
I do not consider it a compromise. It is *worse* than merely
considering the collation interleaving issues on their own
merit.
--Ken
This archive was generated by hypermail 2.1.5 : Fri May 21 2004 - 15:02:18 CDT