just for fun, i tried to build the PDF version of the fabric docs on
my fedora 28 system and, after installing numerous font and
tex-related packages to get around the early errors, i finally ended
up here, with an error message generated trying to process the unicode
characters at the beginning of the architecture chapter:

  ... snip ...

  Chapter 7.
  [189]

  ! Package inputenc Error: Unicode char ⊥ (U+22A5)
  (inputenc)                not set up for use with LaTeX.

  See the inputenc package documentation for explanation.
  Type  H <return>  for immediate help.
   ...

  l.12204 ...quote{N}} contain a special element ⊥
                                                   (empty type), which is
  ?

here is the apparent culprit in arch-deep-dive.rst:

Both ``V`` and ``N`` contain a special element |falsum| (empty type), which is
in case of ``N`` the lowest element. Initially all keys are mapped to
(|falsum|, |falsum|). For ``s(k)=(v,ver)`` we denote ``v`` by ``s(k).value``,
and ``ver`` by ``s(k).version``.

.. |falsum| unicode:: U+22A5
.. |in| unicode:: U+2208

  the HTML seems to be rendered properly, showing the special symbols,
but the PDF generation appears to choke. some online searching led me
here:

https://doughellmann.com/blog/2016/07/13/using-unicode-with-sphinx-restructuredtext-and-pdf-output/

which seems to explain precisely what i'm seeing, but i don't know
enough about messing with sphinx to know what (if anything) needs to
be fixed.

  thoughts? again, this is on a current fedora system to which i've
just added a ton of texlive-* packages to get this far.

rday