At 18.18 +0100 2001-02-19, jbezos wrote:
>> Question raised by this: Can OCPs output control sequences, or do they just
>> produce characters?
>
>They can output any token (IIRC there is a bug
>when \input is used, but I'm not sure). In fact,
>tokens are necessary when translating Unicode to,
>say, OT1.
>
>>Can one specify what catcode the characters should have?
>
>Unfortunately not. The catcodes used are the
>catcodes when the replacement is done. That means
>that "private" names containing @ cannot be
>used (in general, or if \csname is used).

OK, so in fact the OCPs cannot produce tokens (they just output
characters), but that is not really a restriction as long as some character
has catcode 0 and \csname (or some disguise of it) is available. (The
former condition could be a problem in \verb-like contexts. One would
probably have to have a "private escape" character for the OCPs.)

On the other side of things, how does Omega handle "lost character"
conditions? The current TeX behaviour of ignoring the character and
possibly putting an info message in the log file could certainly be
improved ...  When one is typesetting normal text the approriate action
would be (a) substitution with a character from another font (LaTeX can do
this for characters are represented by encoding-specific commands, but
there are no such mechanisms in TeX for explicit character tokens), (b) an
error message, or (c) a combination of the two. When one is typesetting
verbatim (or verbatim-like) text however, the priorities are different. In
particular, I would like to have some recourse to (d): Typeset a suitable
representation (e.g. U+0312, in a suitably distinct font) of the Unicode
for the character.

Lars Hellström