Tom Lane wrote: >>My patch counts inittapes(), tuplesort_begin_heap() and >>tuplesort_begin_index(), and collect them, and sum them through the >>stat collector. > > > Hm, that doesn't seem like quite the right level to be counting at. > Shouldn't you be hacking fd.c to count operations on FD_XACT_TEMPORARY > files?
Why do you think so? I don't see tuplesort.c is good or not. But all code of sort operations are in tuplesort.c. So I thought it is a good place to count them up. Of course, I can move my code to fd.c if it will be reasonable. -- NAGAYASU Satoshi <[EMAIL PROTECTED]> ---------------------------(end of broadcast)--------------------------- TIP 7: don't forget to increase your free space map settings