>     For Issues, as Lassi has noted, the notion of current directory
>     needs to be clarified or refined, or I'm pretty sure mentioned there.
>
> I'm pushing back on this.  Posix says the current directory is
> per-process, and for SRFI 170 purposes, per-process it is.  Implementers
> of SRFI-170 can add change-thread-directory and its setter, or simply
> derogate from the SRFI in their documentation.  I've added a note to
> this effect.

Unfortunately `working-directory` and `set-working-directory` are the
least of our problems. The main problem is every other procedure in the
SRFI that takes a pathname. In Schemes like Gambit and Kawa, it would
most likely confuse users if those pathnames are not interpreted
relative to the thread CWD.