>  Should probably say `a timestamp object'.

Thanks, will be fixed in the next draft.

 > Why TZPATH and not TZDIR?  There might be a good reason for that, but
  it is a shame the library will not work out of the box on GNU Guix
(and other non-FHS systems), so it would be nice to document the rationale.

I was not aware of TZDIR, I will document it. TZPATH is used because it
is was Python uses, there is no strong reason for it. I will note TZDIR
also.

 >  I suspect there is a typo in the `is has'.

You are correct, I will fix it.

 > Equivalents of these C functions are not present, but they are useful.
Is this omission intentional?

The reason they are not supplied is:

1. Scheme doesn't have a well supported standard for querying locale
information. So likely the date and time information will be in English.
2. There isn't a standard formatting system. There are systems that use
Common Lisp-style formats, C-style formats, string interpolation style,
and more out-there stuff like monadic formatting [1].
3. The SRFI allows for non-Gregorian calendars, and the strftime system
really only works for Gregorian-style calendars.

All of the primitives are here to write a strftime function, though
(except for locales).

 > I find it bit unfortunate to reference non-free documents.  It makes
it impossible to implement the SRFI without buying the ISO document.
Maybe it would be better to describe the rules directly in the SRFI?

Yes I will definitely do this. I have not written the rules directly yet
because the draft was entirely done, and I wanted to get it out there
and not have it sit on my computer while I sorted out the drudgery of
acquiring said non-free standard. (That part of the specification and
the sample implementation was not written by me, and I am trusting that
they implemented it correctly.)

[1]: https://srfi.schemers.org/srfi-166/srfi-166.html

-- Peter McGoron