I wrote: > There's something else going on, because I'm still getting the > assertion failure on my Mac with this fix in place. Annoyingly, > it goes away if I compile with -O0, so it's kind of hard to > identify what's going wrong.
No, belay that: I must've got confused about which version I was testing. It's very unclear to me why the undefined reference causes the preceding Assert to misbehave, but that is clearly what's happening. Compiler bug maybe? My Mac has clang 15.0.0, and the unhappy buildfarm members are also late-model clang. Anyway, I did note that the preceding line res = jperOk; is dead code and might as well get removed while you're at it. regards, tom lane