The following review has been posted through the commitfest application:
make installcheck-world: tested, failed
Implements feature: tested, failed
Spec compliant: tested, failed
Documentation: tested, failed
Hi! I've looked into the patch.
There is not so much of the code. The code itself is pretty clear and well documented. I've found just one small typo
"transactionn"in HeapTupleSatisfiesNonVacuumable() comments.
Chosen Snapshot type is not a solution to the author's problem at hand. Though implemented SnapshotNonVacuumable is
closerto rational choice than currently used SnapshotDirty for range sketching. As far as I can see this is what is
get_actual_variable_range()is used for, despite word "actual" in it's name.
The only place where get_actual_variable_range() is used for actual range is preprocessed-out in
get_variable_range():/** XXX It's very tempting to try to use the actual column min and max, if * we can get them
relatively-cheaplywith an index probe. However, since * this function is called many times during join planning, that
could* have unpleasant effects on planning speed. Need more investigation * before enabling this. */
#ifdef NOT_USEDif (get_actual_variable_range(root, vardata, sortop, min, max)) return true;
#endif
I think it would be good if we could have something like "give me actual values, but only if it is on first index
page",like conditional locks. But I think this patch is a step to right direction.
Best regards, Andrey Borodin.