Heikki Linnakangas <heikki(dot)linnakangas(at)enterprisedb(dot)com> wrote:
> I did some further changes, refactoring SkipSerialization so that
> it's hopefully more readable, and added a comment about the
> side-effects. See attached. Let me know if I'm missing something.
I do think the changes improve readability. I don't see anything
missing, but there's something we can drop. Now that you've split
the read and write tests, this part can be dropped from the
SerializationNeededForWrite function:
+
+ /* Check if we have just become "RO-safe". */
+ if (SxactIsROSafe(MySerializableXact))
+ {
+ ReleasePredicateLocks(false);
+ return false;
+ }
If it's doing a write, it can't be a read-only transaction....
-Kevin