ccl decoding unsafe in 21.1

Current `ccl_driver' implementation in "stable" 21.1 release is
unsafe.  That is, ccl program specified in elisp can cause crashes of
xemacs.

More precisely, when called by `mule_decode', it will output strings
of bytes that are not in mule internal coding (further called mic).
Traversal of such a strings by `bytecount_to_charcount' or so may
cause creating invalid byte pointer and attempt to dereference it,
which will cause crash.

Such a strings are generated when ccl code "writes" an 8-bit character
(that is, value between 0x80 and 0xff).  So unsafe programs are all
programs that decode from coding which includes iso-8859-1 characters,
including `ccl-decode-koi8' from `cyrillic.el' in xemacs-21.4.

21.2 branch contains change `2000-10-05  MIYASHITA Hisashi
<[email protected]>' that fixes it.  In that version when `ccl_driver' is
used for decoding to mic, it handles ccl program output safely by
always performing conversion to mic, even on 8-bit characters.

Are there plans to port such a change to 21.1 branch and make ccl
programming in it also crush safe?