groff ms fidelity to AT&T ms behavior (was: Footnote line length ratio to current line length)

[G. Branden Robinson]

Hi Doug!

At 2020-12-07T15:25:02-0500, M Douglas McIlroy wrote:
> > How much fidelity to AT&T ms behavior do people expect from ms?
> > I've begun to surmise that the answer is "a lot less than they
> > expect from man(7) documents".
> 
> Having a 40-year back catalog of -ms documents, I do care. Footnote
> line length doesn't matter much from an aesthetic standpoint. In fact,
> the short length  has always seemed strange. Unfortunately, though,
> documents have often been tweaked to make nice fits to the pages. That
> makes me leery of changes that could alter the line count on a page.

Right.  Incidentally, I said last month[1] that the historical AT&T ms
footnote ratio was 5/6, but this is clearly wrong as my own email
earlier today to this list illustrated with links to TUHS archives of
V6 and V7 sources.

This is one reason I footnote even my emails.  I have found that I
cannot trust my brain to recall facts accurately unless I brandish a
source citation at it, menacingly.  ("Back, hippocampus!  Back!")

I can't find it right now, but while troubleshooting another ms problem
recently I discovered that Heirloom ms and groff ms put different
amounts of space before or after the page header, at least in nroff
mode.  I didn't chase it down at the time because I was hunting other
game, and now I can't locate the message where I noted it.

> But not very leery if the symptom can be fixed by a single .FL.
> 
> As for man(7) documents, authors' macro usage and even the -man
> package itself vary so much that I have no expectation that a
> 40-year-old man page will come out exactly right. But, hey, it's a
> manual. It's the facts that matter, not the appearance. The facts
> usually shine through the incompatibilities. And in extremis, you can
> look at the source, which in the case of -man often exhibits
> logorrhea, but rarely troff arcana.

I'm hard-pressed to counter that perspective.  One of my frustrations
with the way the Unix manuals worked out historically was that man page
sources made it into the distributions but the ms documents that filled
Volume 2 generally did not.  I have sung the praises of the Volume 2
white papers before.  But parts of the Unix culture that overlooked
these works, either due to lack of access, license, or care, seem to
have elevated the man page and only the man page to a status of
exclusive sufficiency as documentation.

My maximalist wish would be for all the historical Volume 2 documents to
be available in source form under a license like CC-BY-SA.  But I don't
know what the copyright story is with them and it could be complex.  I
have surmised that Prentice Hall still has the rights to Ritchie's C
Reference Manual, and won't let go.

My fallback aspiration would be for those documents to be available
under an unlimited-distribution but limited-modification license
(perhaps no modifications allowed at all), for historical interest and
regression testing.

As far as I know there is no curated collection of ms documents against
which ms implementations can be measured.  Are you in a position to
share some of your collection with the community?

[1] https://lists.gnu.org/archive/html/groff/2020-11/msg00089.html

signature.asc
Description: PGP signature