[bug #42675] [troff] \} treated as macro argument

"G. Branden Robinson" <[email protected]> Sun, 7 Jun 2026 14:50:14 -0400 (EDT)

Newsgroups	gmane.comp.printing.groff.bugs
Message-ID	<[email protected]>

Follow-up Comment #16, bug #42675 (group groff):

[comment #15 comment #15:]
> Oops, I didn't read thoroughly enough: comment #8 used erroneous input and
> comment #10 has the corrected output, and it seems to have omitted Solaris 10
> (though which nroff was used for one of the examples is ambiguous).

Right.  Back then I didn't have a copy of Solaris 10 _troff_ compiled for my
local host.

Interested folks can find [https://github.com/n-t-roff/Solaris10-ditroff
Solaris 10 _troff_ at GitHub].

However, in the spirit of information gathering, here's what happens on a real
Solaris 10 host.


$ uname -a
SunOS gcc-solaris10 5.10 Generic_Virtual sun4u sparc SUNW,SPARC-Enterprise
$ cat count-args.roff 
.de CA
\\n(.$
..
.if n \{.CA\}
.if n \{.CA \}
.if n \{.CA foo\}
.if n \{.CA foo \}
.pl \n(nlu
$ nroff ./count-args.roff | gcat -s
1 1 1 2



So that's interesting.  This makes Plan 9 _troff_ less of an outlier.

Maybe something in the guts of the formatter is sensitive to machine
endianness in a way that affects macro argument processing?

Or maybe Carsten Kunze fixed something when updating Solaris 10 troff to build
on modern systems?

Maybe the most that can be said in _groff_ documentation is that various
_troff_s count arguments differently when right-brace escape sequences are
involved.


    _______________________________________________________

Reply to this item at:

  <https://savannah.gnu.org/bugs/?42675>

_______________________________________________
Message sent via Savannah
https://savannah.gnu.org/

signature.asc (application/pgp-signature, 228 B)