Re: Would it be reasonable to list the fonts that are available by default in groff?

[G. Branden Robinson]

At 2021-07-15T12:10:05-0400, T. Kurt Bond wrote:
> Perhaps in groff_font(5) and in groff.texi?

Thanks for raising this point; it is a poorly understood one.  And that
likely means we don't document it well enough.

The available font repertoire is specific to each output device, so
groff_font(5) or the corresponding sections of our Texinfo manual are
not, perhaps, the best place.

It's hard to situate the material where you're looking for it, because a
*roff formatter does not enforce the availability of _any_ font or
style.  (There is always a default, but it is determined by the device
description file, "DESC".)  In practice, documents or macro packages
must to do this, by attempting to select a font/style by name and
risking failure.

In _practice_, especially in the years groff has been around, it seems
it's pretty reasonable to assume the availability of styles R, I, B, and
BI.  And for typesetter devices, the impact of PostScript on the digital
typography industry suggests to me that you can assume that the 13 base
fonts of its original specification[1] will be available everywhere,
especially now that metrically-compatible fonts are available as Free
Software.

Given the coupling between output device and font repertoire, I submit
that the most appropriate place to document the available repertoire is
in the man page for the relevant output driver.  Four of our seven such
pages do in fact cover this material; see grodvi(1), grolbp(1),
gropdf(1), grops(1), and gxditview(1).

grohtml(1) and grotty(1) omit this information semi-understandably,
since they're not in control of font rendering in the same way that
other output drivers are.  If we had/have good CSS support in grohtml,
on the other hand, for it this objection evaporates.  But my
understanding of grohtml is terrible.[2]

It's also probably a good idea for grotty(1) to explicitly enumerate the
fonts (really styles) that it supports, particularly since style BI is
always available--though not always exposed by macro packages, because
support for the Linotron 202 which featured this style was a
late-breaking development in the rather logarithmic trajectory of *roff
development.

There are some deficiencies; some pages, like grolbp(1), don't actually
mention which font family is the default (news flash--it's Times).

grodvi(1), gropdf(1), and grops(1) list and describe their available
fonts, and when rendered with themselves, setting them in their own
typeface.  It's a nice effect and one I have preserved despite my broad
agreement with Ingo and others that man pages should be written with
portability foremost in mind.

Moreover, to your overall point, our documentation could make the point
in my first paragraph more clear.

As always, I welcome correction on any point above.

Just as a heads-up, I happen to have been working on groff_font(5) and
its corresponding Texinfo nodes today, but I am not yet tackling the
deficiencies I've noted above.  So fixes for them won't necessarily be
in the next push (groff-commit list traffic) you see.

Regards,
Branden

[1] Take the Cartesian product of Times, Helvetica, and Courier (a
    serif, sans-serif, monospaced family, respectively) with the 4
    styles noted above--that gets you 12 faces.  The "Symbol" font is
    the 13th.

[2] To my delight, Gaius Mulley recently claimed a Savannah ticket to
    work on, <https://savannah.gnu.org/bugs/?60782>.

signature.asc
Description: PGP signature