There are several largely disjoint sub-problems tackled by this SRFI:

- Processes, signals, file descriptors, user and group IDs
- File system (i.e. dealing with files by name instead of fd/port)
- Glob syntax (could be implemented in fully portable Scheme)
- User and group information (getpwent() et.al.)
- Text terminal facilities (tty, curses, Windows console)
- Networking and sockets (currently out of scope for this SRFI)

Are we being overambitious? Should we partition things into multiple SRFIs?