Thanks as always,Yong Huang.
"But 10053 doesn't require actually running the SQL. The v$ view is populated after the SQL is run"
So is it true that the values in v$ view are always more convincing than in 10053?
And here is another issue I think I might need your comment:
I did some tests on hash join cases these days, if I used "set autot trace exp" before the sql was run, I found that the plan output had the column "TempSpc", so I thought this was a onepass execution at least(it can not finish its job within the pga);but when I checked the LAST_EXECUTION column in v$sql_workarea after the sql run, it said "optimal" and the corresponding LAST_TEMPSEG_SIZE column had no value.I was confused with that.
In this case, who I should believe?The "autot trace exp" output or the v$sql_workarea?
Thanks in advance!
[ 本帖最后由 sprite731 于 2011-6-16 22:47 编辑 ]
|