decoders writing control-1 chars?

[stephen at xemacs.org]

Ilya N. Golubev writes:

 > > CCL is underdocumented and not easy to test

 > So better fix that.

Are you volunteering?

IMO, the XEmacs project has no need for CCL.  If some of our users are
using it, fine, we can keep it.

XEmacs project resources would be better spent on converting current
usage of CCL to table lookups.

 > CCL coding systems are likely to remain for quite a bit, even in
 > 21.5.

With exception of Mule-UCS, all CCL coding systems I know of can be
trivially coded in C.  They're just 256-entry table lookups, and we
already have tables for most of them.

If you know of any IANA-registered coding systems currently
implemented for Emacsen in CCL that cannot be converted to a 256-entry
table lookup, please let me know, and I'll reconsider.  But Mule-UCS
is already obsolete, the rest can't be very hard.