Re: CHECKSTYLE suggestions: unnecessary quotations and unnecessary \f escape

[G. Branden Robinson]

At 2022-03-20T13:57:18+0100, Ingo Schwarze wrote:
> To summarize, i certainly wouldn't want to warn about these
> mini-issues in mandoc -T lint.  Admittedly, warnings in groff accept a
> significantly higher rate of noise and false positives than in mandoc
> -T lint,

Excuse me?  If you are referring to the groff man(7) `CHECKSTYLE`
feature, you are throwing a gauntlet, sir.

None of the checks that key off of that register are heuristic in any
way.  They are strictly defined and emitted only when their conditions
obtain.  If this statement is incorrect, I would like to know so that I
can fix it at once.  So far in my experience they have been absolutely
reliable[1].

Which warnings did you have in mind?  If you mean the yammering about
nodes and transparent throughput and whatnot, I have news for you[2].

Unfortunately, grohtml is still loud as hell if one dares to include
tables or equations.  If I ever understand it well enough to judge
whether its diagnostic noise about "suppression boundaries spanning more
than a page" can be silenced with low risk to the diligent user, I will
do so.  But there are so many problems with grohtml (try "groff -Thtml
-me meref.me", for instance), that it is hard to know where to start.

On the bright side, we don't run "groff -Thtml" much as part of a build.
I see the ghastly spew because I regularly generate HTML versions of all
our man pages, to catch regressions.

> but i'm not really sure about the tradeoff of noise vs. usefulness
> for these two even in groff.

I've been getting groff's own build quieter and quieter.  That main
sources of diagnostics from groff itself now are the missing glyphs in
the Base 35 Type 1 fonts, about which I posted last week[3].  If we can
crack that nut we'll be down to one.

There's another Type 1 font problem--about PostScript glyph names
"Upsilon" and "Upsilon1" both being mapped to the groff glyph name
(special character identifier) "*U".  This comes from the file
font/devps/symbolmap in the source tree.  I keep meaning to do something
about it, but I don't have a canonical dump of a vintage Adobe Type 1
font to see which one looks like a Latin capital Y and which is
distinguishable by a hook on the right arm (or indeed, if they differ
depending on the font in question).  If they're truly the same, I guess
we can just drop the mapping for "Upsilon1".

Does someone here know?  Does someone have a real Adobe Type 1 Symbol
font they can pop open in FontForge and take a screenshot of?

Regards,
Branden

[1] This claim surrenders a Beslan school full of hostages to fortune.
[2] 
https://git.savannah.gnu.org/cgit/groff.git/commit/?id=557bc0558dfdee7e3f2011433cf4606052e4e7e1
[3] https://lists.gnu.org/archive/html/groff/2022-03/msg00016.html

signature.asc
Description: PGP signature