"Magnus Hagander" <mha@sollentuna.net> writes:
>> I could vote for:
>>
>> bool pg_query_cancel(int)
>>
>> backend_cancel or cancel_backend sounds like it should
>> terminate the entire backend like kill -TERM would do.
> IIRC, the original discussion had a possible pg_query_cancel(int)
> functoin that would cancel a query based on XID or something like that,
> and we wanted to differentiate from that. (No such function was ever
> created, but it was the reason, IIRC)
Right. Rod's suggestion is superficially more logical, but it doesn't
fit into the plans for future extension of the capability.
regards, tom lane