> Doing nothing for calls now is fine by me, though I > agree that this could be improved at some point, as seeing only 1 > rather than N for each fetch depending on the size is a bit confusing.
I think we will need to clearly define what "calls" is. Perhaps as mentioned above, we may need separate counters for "calls" vs "fetches". This is definitely a separate thread. > Doesn't this comment at the top of ExecutorRun() need an update? It > seems to me that this comment should mention both es_total_processed Yes, updated in v5. > There is no need for this part in ExecutorFinish(), actually, as long > as we always increment es_total_processed at the end ExecutorRun() for > all the operation types? Ah, correct. I changed that and tested again. > - es_processed: number of tuples processed during one ExecutorRun() > call. > - es_total_processed: total number of tuples aggregated across all > ExecutorRun() calls. I thought hard about this point and for some reason I did not want to mention ExecutorRun in the comment. But, I agree with what you suggest. It's more clear as to the intention of the fields. Attached is v5 addressing the comments. Regards, Sami Imseih Amazon Web Services (AWS)
v5-0001-Fix-row-tracking-in-pg_stat_statements.patch
Description: v5-0001-Fix-row-tracking-in-pg_stat_statements.patch