Back when this ticket was about fixing a syntax error, before it morphed to an existential question, Ingo opined:

comment #1:

> If there is value in retaining the file at all, then it is in
> retaining it with as few changes as possible.

Branden countered:

comment #2:

> My preference would be to re-open the ticket and fix it.

Ingo's desire for it to be a file in amber is already undermined by the handful of trivial fixes it's garnered over its groff/git lifetime.  Fixing the legitimate syntax error Bjarni pointed out seems to be in line with these changes.

...insofar as removal of obsolete code is a "New feature."  It's the best fit of the available Item Groups, though.

comment #5:

> I don't have an opinion on doc-old.tmac in particular, but I agree with the general sentiment that code distributed with modern groff releases should be intended to be used.  So any code not fitting that expectation could reasonably be removed.

Thanks, Dave.

That's three votes in favor of mothballing it, all from regular development list participants.  That's a roaring consensus by our sleepy standards.
 

> If appropriate, maybe a line in the release notes could explain how to retrieve the last version of the file from git, or whatever archive seems most definitive.

Yes; I think noting it in the NEWS file ("for reconstruction of historical documents") could point people to the groff 1.22.4 release archive for that--that's going to be readily available for many years, judging by <https://ftp.gnu.org/gnu/groff/old/>.  (Even the non-"old" release archives date back to 2009.)

Ingo, would you like to do the removal honors?  I can field the NEWS thing.

  Instead of removing the line  "..",
make it a comment line with an explanation,
something like:

.\" The original line '..' was commented out
.\" to avoid a warning about an
.\" undefined macro,
.\" and to preserve it for historical reasons.

I'm not sure I agree with this being closed.

My opinion:

The GNU roff codebase is not supposed to be a museum.

Our macro packages should exhibit exemplary style (literally, serve as good examples) insofar as we are aware and able to make them.

If there is more value in keeping mdoc-old frozen as an exhibit than in serving as macro package for formatting documents even if the user has the warnings dialed way the hell up, then I submit that it would be better to drop it from the groff distribution.  Let other implementations be in the archaeological exhibit business.  Certainly there are plenty of moribund SourceForge, GitHub, and even Apache Software Foundation projects (<cough>Apache Open Office</cough>) fitting this description.

My preference would be to re-open the ticket and fix it.

Ingo and I have weighed in; would anyone else like to?

From 3b96b21c1fead1d92eb658bffe25411bb800fa9d Mon Sep 17 00:00:00 2001
From: Bjarni Ingi Gislason <bjarniig@rhi.hi.is>
Date: Fri, 17 Jan 2020 19:19:49 +0000
Subject: [PATCH] doc-old.tmac-u: Remove a superfluous "end of a macro" (..)

  "nroff -ww doc-old.tmac-u" reports on standard error:

troff: doc-old.tmac-u:273: warning: macro '.' not defined

Signed-off-by: Bjarni Ingi Gislason <bjarniig@rhi.hi.is>
---
 tmac/doc-old.tmac-u | 1 -
 1 file changed, 1 deletion(-)

diff --git a/tmac/doc-old.tmac-u b/tmac/doc-old.tmac-u
index ddb782f7..ef597d97 100644
--- a/tmac/doc-old.tmac-u
+++ b/tmac/doc-old.tmac-u
@@ -270,7 +270,6 @@
 .if n \{\
 .ad l
 .na
-..
 .\}
 .\" PREDEFINED STRINGS
 .if t \{\
--
2.24.1