From: Shriram Krishnamurthi <xxxxxx@cs.rice.edu>
   Date: Sun, 21 Nov 1999 18:21:11 -0600 (CST)

   I am a little concerned that there has been no discussion about Oleg's
   updated SRFI document.  I would appreciate it if Richard could comment
   on whether he thought the revisions are in line with his views -- the
   text is clearly oriented along his suggestions.  It would be nice to
   get feedback from other people, too -- so far, only one person has
   provided substantial comments.

The new draft is much easier to understand than the original.
I expect I will implement it in Scheme 48.

I do have one suggestion.  The SRFI says:

  An implementation is however encouraged to treat an erroneous or
  unrecognizable <hash-comma-datum> as a compound datum
  (<datum> <datum>*) -- that is, pretend that the "#,(" token was a "("
  token. This means that the read function shifts the burden of dealing
  with an erroneous #,() form to its caller.

This is a very bad idea.  Implementators should be encouraged to
report errors, not to silently proceed as if nothing has happened.
Treating `#(' as `(' doesn't shift the burden of dealing with an
error, it hides the fact that there ever was one.  How can the caller
deal with an error that it can no longer detect?

                                          -Richard Kelsey