I wrote:
> Hm. We actually document PQprint. While the other two entry points
> in that file aren't documented, they are listed in exports.txt,
> which means somebody *could* be calling them. I think we'd at least
> have to go through a long deprecation period before we could consider
> removing them.
Meanwhile, returning to the point at hand in this thread: AFAICS
from a quick review of the git logs, the last intentional changes
of user-visible behavior in fe-print.c were more than 20 years ago.
Since then it's only seen in-passing bug fixes, such as libpq-wide
fixes of signal handling.
So, to the extent that this code has any use at all it's for
legacy applications. I don't think we should be making random
changes in its behavior for at-best-weak reasons.
In short, I'm now a vote to revert every single bit of 5da9868ed.
regards, tom lane