[racket] syntax, differently

From: Shriram Krishnamurthi (sk at cs.brown.edu)
Date: Wed Aug 18 15:17:00 EDT 2010

I've wanted to see . liberated for a long time.

I agree that the shift from PLT Scheme to Racket opens up this
possibility.  Fortunately, we don't already use . for much already.  I
suspect the number of actual uses in source are very small.

(Indeed, it opens up the possibility that a whole bunch of characters
that could be, but aren't actually, used in identifiers be free for
consumption.  This is especially reasonable given that we have been
idiomatically shifting away from symbols to strings, so symbols no
longer need to carry string-like burdens as they did in Lisp days --
with '|...| still around when necessary.)

However, Matthew gets to make this decision, and I'm sure whatever he
decides will consider a half-dozen angles that the rest of us didn't
know even existed.

Note that the overall issue remains somewhat non-trivial, because a
"floating" . already does duty in lists, AND does additional duty as
the infix operator designator, and furthermore this would be a stretch
from what Racket currently does: a.b.c might be simple enough, but
(foo x).b.c is a bit farther afield.


Posted on the users mailing list.