> I think what happens then is that SRFI-170 procedures work with the
> per-process $CWD and the R7RS procedures work with the per-thread $CWD.
> I agree that this is confusing, but I don't know what to do about it all.

That's my interpretation as well.

Could we simply add some prose to SRFI 170 permitting implementations to
translate relative pathnames per the thread CWD before passing them to
the syscalls? While at it, we could also allow translating pathname
objects to strings (for Schemes like Kawa that have a path datatype).