Per Bothner:
> But I can't see why you'd want to accidentally invoke nfx,
> which will happen if it's in the "user's namespace" (to use C/C++
> terminology).  If somebody wants to re-bind the meaning of
> c-expressions, then redefining a $nfx$ macro is easy enough, but don't
> hijack an identifier the user might be using innocently.

John Cowan also prefers an "unusual" macro.  This has been "nfx" for years, and more user-visible than the bracket stuff, so if its name changes at all, "$nfx$" would be better (as it'd be less of a change, while still making it "highly unlikely to be accidental").

This would be a broad, beyond-Scheme change, so I'll need to bring this up on the readable-discuss list.

--- David A. Wheeler