Alvaro Herrera <alvherre@commandprompt.com> writes:
> Hmm, would there be a problem if a scan on catalog A yields results from
> supposedly-running transaction X but another scan on catalog B yields
> result from transaction Y? (X != Y) For example, a scan on pg_class
> says that there are N triggers but scanning pg_trigger says N-1?
Yeah, I came to that same conclusion downthread.
regards, tom lane