Hello, Andy Wingo <wi...@pobox.com> writes:
> Just a random thought: we'll have to deal with unicode ports, and the > byte-order-mark, and read-char consuming more than a byte. But perhaps > that work can come after Guile actually supports UTF-32 internally. R6RS > is probably a good guide in this regard, cleanly distinguishing bytes > from characters, and specifying the codecs for ports. Yes, definitely. The `(rnrs io ports)' R6RS library has the notion of "codecs", so it may be that we should just follow this design. However, it will probably be hard to provided a compatibility wrapper for our current port "API" such that existing third-party port implementations can be recompiled without any change. Thanks, Ludo'.
