I've just posted on the mailing list a revision for SRFI-105:
* The marker is now #!curly-infix, not #!srfi-105.
* It's documented that the default environment shouldn't define nfx (other than maybe an error).  People can important an nfx implementation, but now it'll be obvious where that occurs in the source code.
* The code now includes a complete implementation, not just the key routines.  You can copy & paste the code into a file, run it, and have a preprocessor that translates c-expressions into s-expressions.
I hope that the SRFI editors will get a chance to update the posted version soon.

So... are we done?

The key thing, really, is adoption.  There's no point in writing a SRFI unless it'll be adopted by Scheme implementations and users.  So, have we maximized the likelihood of adoption in Scheme implementations and users?  If not, what do you think should be changed to maximize adoption?

--- David A. Wheeler