On Thu, May 04, 2023 at 08:39:03AM -0400, Tom Lane wrote: > Hmm, I'm not sure why PLy_trigger_build_args's pltargs needs to > gain a "volatile" here? LGTM otherwise.
I removed that new "volatile" marker before committing. I was trying to future-proof a bit and follow elog.h's recommendation to the letter, but following your mental model upthread, it doesn't seem to be strictly necessary, and we'd need to set pltargs to NULL after decrementing its reference count in the PG_TRY section for such future-proofing to be effective, anyway.
