OK.. Dave Goldsmith is mostly right - if I follow his logic,
it should be feasible to get ISO10646 set up as a MIME charset.
If I follow Masataka Ohta's logic, he's correct as well -
ISO10646 botches a few things in a less-than-optimal manner.
Is it however a mortal sin to sleep with the devil and do this:
1) All parties concerned agree that ISO10646 is flawed, work
out the needed language to do damage control and specify how
best to use the parts that work, and register the charset.
Perhaps we need a disclaimer section that discusses known
weaknesses in 10646?
2) Continue applying pressure to ISO to Get It Right this time,
and when son-of-10646 comes out, register that charset in preference.
Yes, 10646 is less-than-optimal. But the same can be said for
8859-1, which covers even fewer languages correctly but doesn't
seem to cause major fits...
Computer Systems Engineer