[r6rs-discuss] Stateful codecs and inefficient transcoding

[William D Clinger]

Sheesh.  Let me try that a third time:

 *  If t1 and t2 are transcoders, then their composition
    is defined by describing their composition in both
    the input and output directions.  In the input
    direction, their composition is t1input followed by
    t1output followed by t2input.  For output, their
    composition is t2output followed by t2input followed
    by t1output.

So on input, you read with the old transcoder, undo the
read with the old transcoder, and then read with the new.

On output, you write with the new, undo the write with
the new, and write with the old.  That will usually be the
same as writing with the old, while limiting the character
set to the new.  (I'm not sure that's the Right Thing, but
switching transcoders on output doesn't seem to be a real
requirement anyway.)

I apologize for the mess, by which I mean more than just
these three messages.

Will