Vik Fearing <vik@postgresfriends.org> writes:
> #18776 appears to have been discarded as a duplicate. In any case, no
> one is going to fix a bug in 9.4 which has been EOL since 2020. I
> suggest you upgrade to a supported version and re-report the bug if it
> still exists.
If you do reproduce your problem on a still-in-support branch,
please re-file with a *self-contained* test case. It's impossible
for someone to reverse-engineer your problem from just the two
function definitions, with no table declarations, sample data,
calling sequence, or expected vs. actual results.
https://wiki.postgresql.org/wiki/Guide_to_reporting_problems
If it's difficult to cram all the details into our bug-reporting
form, personally I'd ignore the form and just send mail to
pgsql-bugs@lists.postgresql.org
regards, tom lane