Could we do some kind of semi-formal push to come to an agreement on the
remaining language issues on which Scheme implementations have split
into opposing camps? There shouldn't be that many:

* Keywords and keyword arguments
* Pathname representation
* (Classes/methods?)
* (Probably something else?)

I've been gathering energy to finish SRFI 177 (keyword arguments). These
fundamental language discussions are probably a bit fatiguing to all of
us, but it'd be encouraging if we can make a list of the remaining
issues. Then we can estimate how much work is ahead of us. Any decisions
we make ought to also be very useful information for planning the rest
of the R7RS-large effort.

The point would not be to force one solution to each issue on which
there is a split of opinion, but to find enough common ground that
portable libraries and SRFIs can be written. For each issue we could
define a minimal portable interface that interoperates with each side of
the split. Then maybe write a SRFI about it (in a similar vein as SRFI
177 and some of John's SRFIs now handle such compromises). Future
interoperable libraries can then depend on that interface and hence have
the option of eventually working on most Scheme implementations.