I've submitted to the SRFI editors an updated SRFI-110, here:
  http://srfi.schemers.org/srfi-110/

This includes a lot of changes due to my taskmaster Mark H Weaver :-).
In particular, it supports sweet-expression datum comments and
a less-complex grammar that is still rigorous, as well as a variety
of small changes.

Sweet-expression datum comments are
started by "#;" followed by whitespace in the first position
(exactly the same as abbreviations).
This required adding an "empty" return value, which meant that
I needed to define and use several procedures to handle empty.
On the other hand, now that "empty" exists, a few rules could be
simplified.

Mark H Weaver commented about the grammar complexity.
I haven't (yet?) found a significantly-better way to notate the
grammar that's not LL(1).  *However*, I've done a number
of modifications to make the key BNF grammar simpler.

Does the "key BNF" grammar look like an improvement?

--- David A. Wheeler