[Q21.5] Add support for non-ISO2022 8 bit fixed-width coding-systems.

[Stephen J. Turnbull]

QUERY 21.5

Please clarify some points.

Aidan Kehoe writes:

 > 	Provide make-8-bit-coding-system, a way to construct a fixed-width
 > 	XEmacs CCL coding system given a map from 8-bit characters to
 > 	Unicode. Such generated CCL coding systems do not have the Latin
 > 	deunification problem of iso-8859-15, for example.

I don't understand.  ISO-8859-15 by itself has no "Latin deunification
problem."  Trying to mix ISO-8859-15 with ISO-8859-1 is what causes
the problem, and presumably mixing ISO-8859-5 with a "generated CCL
coding system" will give the same problem.

I can see that 8-bit coding systems that somehow map to Unicode will
be more compatible with each other, and that's an improvement.  But
please document the corner cases more carefully than this.

 > 	Move to accepting characters as well as integers

What do you mean by "as well as"?  If something is semantically a
character, only characters should be accepted, and vice versa for
semantic integers.  If you just mean you're making this distinction
properly, hurray!  But we should avoid polymorphism in these functions
if at all possible.