So many groff discussions get re-routed to TUHS thanks to Doug's MUA. :) At 2021-08-02T19:38:42-0400, Douglas McIlroy wrote: This would tighten the existing semantics of %O, would it not? Maybe that's not a problem, since it's already defined as a dumping ground[1]. So if I understand you and the GNU refer(1) manual correctly, %O would be parsed for a pair of brackets[2] containing keywords only (no fields), and if that resolved to a single reference, the label for that reference would be interpolated. That of itself would be sufficient to halt the recursion. Here's an example of what this sort of thing looks like today, _without_ adding this feature to %O[3]. $ cat EXPERIMENTS/my-db-file-for-doug %A Daniel P.\& Friedman %A Matthias Felleisen %C Cambridge, Massachusetts %D 1996 %I The MIT Press %T The Little Schemer, Fourth Edition %O Reviewed in [posterior analytics 36] $ REFER=EXPERIMENTS/my-db-file-for-doug ./build/test-groff -R -Tutf8 \ -ms EXPERIMENTS/scheme.ms | cat -s Read the book[1] on your summer vacation. Commentary is available.* ─────────── [1] Daniel P. Friedman and Matthias Felleisen, The Little Schemer, Fourth Edition, The MIT Press, Cam‐ bridge, Massachusetts (1996). Reviewed in [posterior analytics 36]. * Space reserved for penetrating insight. Regards, Branden [1] "Other information. This is usually printed at the end of the reference." [2] could be anything non-keywordy, but brackets seem like a good choice [3] ...and a nickel goes into the "cat -v swear jar".