Re: [groff] Regularize (sub)section cross references.

[John Gardner]

> or perhaps it was considered but consciously rejected because
> situations existed where it would be detrimental?

It could have been to make sections more conspicuous in text editors,
making it easier for the author to spot sections when skimming a file.
Doubt there was any syntax highlighting back then... =)

Plus it's easier to search for a section name if you know in advance it'll
always be in uppercase, which I guess spares you the effort of remembering
to do a case-sensitive search in vi or whatever...

On Mon, 17 Dec 2018 at 06:59, Ingo Schwarze <address@hidden> wrote:

> Hi Tadziu,
>
> Tadziu Hoffmann wrote on Sun, Dec 16, 2018 at 07:08:06PM +0100:
> > Ingo Schwarze wrote:
>
> >> Jerry Saltzer's RUNOFF (1964) did not have a .tr request, [...]
> >> By AT&T Version 3 UNIX (1973), nroff(1) did have it, [...]
>
> > Doug McIlroy's GECOS runoff had it
>
> Interesting.
> Oh yes, now that you point me to it,
> i see it in the source code.
>
> > (the memo from 1969
> > even mentions its utility in creating unpaddable spaces),
> > so I assumed (perhaps wrongly) that this feature had also
> > been adopted in Unix nroff from the beginning.
>
> That does seem likely, then.
>
> >> so it's not clear to me whether unix v1 roff(1) already
> >> supported the .tr request.  But even if it did, that
> >> would not have helped: programmable macros only appeared
> >> in nroff(1), in 1972, at least a year after the v1 manual
> >> pages were completed - let alone begun.
>
> > I see.  So while the v3 nroff manual page already mentions
> > macros (and the .tr request), the manual page itself was
> > still written using raw nroff requests.  For v4 the manpages
> > *were* rewritten to use macros (so it's not like editing all
> > the manpages was considered too much work at the time),
>
> Major formatting changes occurred at least three times:
>
>  * AT&T Version 4 Unix: use of the first macro sets:
>    man0/naa (74 lines) and man0/taa (91 lines).
>  * AT&T Version 7 Unix: use of the new man(7) macros (267 lines)
>  * 4.4BSD: use of the new mdoc(7) macros (4620 lines)
>
> Approximate numbers of manual pages:
>
>  * AT&T Version 3 Unix (1973):  220 13k lines 53k words 305kB
>  * AT&T Version 4 Unix (1973):  262
>  * AT&T Version 6 Unix (1975):  253
>  * AT&T Version 7 Unix (1979):  301
>  * 3BSD                (1980):  401
>  * 4.0BSD:             (1980):  438
>  * 4.1BSD:             (1981):  439
>  * 4.2BSD:             (1983):  688
>  * 4.3BSD-Tahoe:       (1988):  966
>  * 4.3BSD-Reno:        (1990): 1220 175k lines 840k words 5200kB
>  * 4.4BSD:             (1993): 1461
>  * 4.4BSD-Lite1:       (1994): 1178
>  * 4.4BSD-Lite2:       (1995): 1199
>  * OpenBSD-current:    (2018): 3387 850k lines 4400k words 27MB
>                                     (base only, without X and ports)
>
> So it is hundred times the volume now compared to v4,
> or five times the volume that Cynthia had to handle when
> she did her rewrite.  And i doubt that OpenBSD is the largest
> system out there.
>
> Of course, in the case at hand, we are only talking about the .Sh/.SH
> lines; still, a quick grep gives me 23k of those in the OpenBSD
> base system alone - more than the complete text of the v4 manuals.
> Some of that can certainly be automated, but it still needs
> proofreading, and the automation is not completely trivial.
>
> > but auto-capitalization for section headers was not considered --
> > or perhaps it was considered but consciously rejected because
> > situations existed where it would be detrimental?
>
> Doug or Ken may know that; i don't.
>
> Yours,
>   Ingo
>
>