Erik,
my formulation of the first rule was rotten.
The situation (quite common in Norway :-( ) that the rule was intended
to OUTLAW, was the situation
The character { means a left curly bracket if what you are looking at
looks like C code, but if it occurs in a word like "kj{rlighet"
(love in Norwegian :-), it means the lower case AE ligature.
In this case, it is nearly impossible to write anything less than a
complete AI system with language knowledge to display the message
properly.
New version of the first rule:
The character indicated by any byte or set of bytes must be completely
and unambiguously specified by the registration of the character set,
either by a complete listing of possible characters, or references to
other published documents that contain such listings.
ISO-2022-JP is such a registration (that references other documents).
Better?
Harald A