Re: Changing section header autogenerated by 'refer'?

[G. Branden Robinson]

At 2020-11-14T17:47:01+0100, Johann Höchtl wrote:
> On 11.11.20 23:00, Damian McGuckin wrote:
> > That said, I think the whole 'refer' package and how it fits into
> > the process of producing a document could do with some further
> > explanation.  There is some in the context of MoM but it needs to be
> > generic.
> 
> Yes to that. Refer feels like black magic.

For groff 1.23.0, there will be reference to Lesk's original refer(1)
paper in the "See also" sections of all related groff man pages.

Unfortunately, from what I recall of the paper right now, the
implementation has changed here more than anywhere else, and I think a
lot of that happened in the 1980s before James Clark undertook his
clean-room rewrites, and (2) Lesk spent a lot of time in the paper
considering performance an implementation details that are wholly
irrelevant today thanks to Moore's Law.

However, it's better than nothing, and it's how I got my head around
what the heck it is that refer(1) and its supporting tools do.

lkbib(1) and lookbib(1), for one thing (two things?) are vastly more
general than their descriptions imply.  You can, and I have, used them
for ad hoc test databases of little factoids and miscellaneous
information.  They're like a multi-line grep, and return you a whole
"paragraph" instead of just a matching line.  It's great.  "fortune -m"
works similarly, on an existing database.

It just _happens_ that you can use these for refer(1) databases by
making the paragraphs a series of %A, %B, ..., lines.

If I'm not mistaken, Research-era AT&T Unix had a look(1) command which
was explicitly for quick-and-dirty text database application.

I expect a Murray Hill veteran or Plan 9 maven could say more.

Regards,
Branden

signature.asc
Description: PGP signature