In my rush to finish my overly long message last night, I made some absurd
comments at the end about Richard Kelsey's proposal.  I meant to say that
the READER-SYNTAX form was orthogonal to the rest of that proposal and
could go into another SRFI, but I got carried away and said that the FILES
and CODE forms are orthogonal also.  This is nonsense, because Kelsey was
proposing a separate configuration language, which must be connected to
Scheme code in some way.  I apologize for broadcasting this nonsense.

To clarify further:  I actually like the idea of a separate configuration
language for Scheme programs.  This would make it easier for a compiler to
get around some of the barriers to compiler optimization that are created
by compiling individual forms or files separately.

Will