Paul J Stevens wrote:
Matthew O'Connor wrote:
Quick Summary - I suggest we add the following two indexes:
CREATE INDEX dbmail_headervalue_physmsg_id on dbmail_headervalue
(physmessage_id );
ok. Though I assumed this was implied by the unique key already present.
I think you are referring to:
"dbmail_headervalue_1" UNIQUE, btree (physmessage_id, id)
which is a combination unique key between physmessage_id and id. This
won't apply to all queries that need to join on or contrain on
physmessage_id.
CREATE INDEX dbmail_headername_lower_headername on
dbmail_headername(lower(headername));
I'll changing the unique key on headername to a unique key on
lower(headername)
OK, I thought about that, but I wasn't sure if it was legal, glad to
hear it is.
Matt
