Nice timing on the reverse pull (https://github.com/johnwcowan/srfi-170/pull/7). Besides the things we're cogitating on, the only item left in my specific list was the issue of making clear we were supporting "kill". And it just occurred to me, while I add the canonical POSIX system call, why not append the (1) or (8) command, if any, after the (2) or (3) system call or library call? And list them with the man section, the easiest reference a user will have, like:

(signal-process proc sig) → undefined (procedure) (POSIX signal(2), kill(1))

The time functions are a major exception to the functions that use return record with named field accessors, which also disallows a timespec? predicate. This hit me over the head as I was creating a Chibi Scheme wrapper for clock_gettime, which takes a pointer to a timespec struct. Finishing the whole time section will be even more trivial if we change timespec to the record(?) type the Chibi Scheme FFI creates, not having to copy the contents of that struct into a pair will be cleaner and faster, and faster can matter for something returning time in a granularity that's possibly nanoseconds.

My Chibi Scheme fork is as follows, this URL specifying the srif-170 branch: https://github.com/hga/chibi-scheme/tree/srfi-170, but you might want to check if there's srfi-170-new working branch. As I go through the Implementation for scsh (skipping the creating processes section for now), I'll be creating a Chibi Scheme version that so far also works on OpenBSD, prioritizing what scsh doesn't implement, or significantly differs like the 0.7 file-mode vs. integer mess.