> Isn't that what statement_timeout is for? Since this is entirely based
> on estimates, using arbitrary fuzzy numbers for this seems fine to me;
> precision isn't really the goal.
There's an important difference to statement_timeout: this proposal
would avoid completely taking any resources if it estimates it can't be
executed in proper time, but statement_timeout will allow a bad query to
run at least statement_timeout long...
Cheers,
Csaba.