🚀 go-pugleaf

In article <1993Jan23.214359.29186@porthos.cc.bellcore.com> tony2@prefect.cc.bellcore.com (gozdz,antoni s) writes:
|I use extensively automatic reference processing
|in troff/Eroff. However, the standard implementation
|allows one to use only a two-character string to name the
|reference (I think that's the way troff was set up to
|recognize the \*( sequence, right?). 

|I'd like to use longer, more descriptive strings,
|e.g., \*(weber91a, whose length could be either
|constant (at least 6 characters) or length. 

|Another, related question: is there an internal limit
|in troff to the number of references a single document
|may have (like the number or registers etc.)?

Most versions of troff (probably including eroff) only support 2
character string names, so you're basically out of luck using the .RS/.RF
mechanism.

However, you can use "refer", which essentially bypasses all these
limits, and allows you to say something like [weber91] in your document.
"refer" is a standard part of DWB, don't know about eroff.  If you
don't have refer, you could probably pick up "grefer", which is
a part of groff.  While you need C++ to compile groff, you may be
able to obtain a pre-compiled binary of grefer for your system.
-- 
Chris Lewis; clewis@ferret.ocunix.on.ca; Phone: Canada 613 832-0541
Psroff 3.0 info: psroff-request@ferret.ocunix.on.ca
Ferret list: ferret-request@ferret.ocunix.on.ca

Correspondent would like "long register names."
Sorry, but the "twoness" of troff is pervasive.
There are 96 printing characters (not all of which
can appear in string names; eg I would't try backslash)
and you can use one or two char names, so 96*96+96
appears to be the limit.
The way most people solve this is to append to
a diversinon with your own personal "magic cookie"
and then use a companion awk program.
For example: ***REF1 <contents of reference 1>
Of course, this means you need 2 passes, but you
always need two passes due to the potential for
forward referencing. You then insert your simple
awk program between the 2 passes.

In article <24703@alice.att.com> npn@alice.att.com (Nils-Peter Nelson) writes:
>
>
>Correspondent would like "long register names."
>Sorry, but the "twoness" of troff is pervasive.
>There are 96 printing characters (not all of which
>can appear in string names; eg I would't try backslash)
>and you can use one or two char names, so 96*96+96
>appears to be the limit.
>The way most people solve this is to append to
>a diversinon with your own personal "magic cookie"
>and then use a companion awk program.
>For example: ***REF1 <contents of reference 1>
>Of course, this means you need 2 passes, but you
>always need two passes due to the potential for
>forward referencing. You then insert your simple
>awk program between the 2 passes.

I did not complain that the number of combinations that
could be used in the .RS/.RF pairs is too small;
these two-character strings are just difficult to 
remember when needed again for re-referencing.

The solution offered might do the job, but obviously, 
it's not as painless as it might/should be.

A related question: what other structures in troff use
the \*( prefix?  If it's only the \*(Rf/.RS/.RF
construct, would it be difficult to assign variable-length
strings to it? (just a wild guess; I'm not a programmer). 

Tony
tony2@cc.bellcore.com

Tony Gozdz asks:
A related question: what other structures in troff use
the \*( prefix?  If it's only the \*(Rf/.RS/.RF
construct, would it be difficult to assign variable-length
strings to it? (just a wild guess; I'm not a programmer). 

==
\*( means gobble up exactly the next two characters
and treat it as the name of a string. Strings may be
used by any of the macros: footnotes, references,
table of contents, headers, footers, etc.
\*x, where x is *not* ( means gobble up exactly 1 char.
If string names could be more than 2 chars, about a
million troff documents would stop working, because
folks expect \(abcd to produce "string ab"cd,
not "string abcd".
The syntax was a poor choice, but it is so embedded in
the culture it can't be changed. More modern "languages"
use start and end delimiters, rather than fixed length
tokens, although this does lead to verbosity
(as in \stringname abcd \endstringname).