"Hiroshi Inoue" <inoue@tpf.co.jp> writes:
>> Remove the 'index deactivation' code, since it provides
>> no useful protection in the shared- index case.
> Seems a funny reason to me.
Well, as you know I never liked that code; modifying permanent on-disk
data didn't seem like a sensible way to protect against intra-transaction
interlock problems. So when I saw a chance to get rid of it, I thought
we should take it.
> I've spent a pretty much time only to put back your change.
Why? As far as I know, the modified code does everything you want.
regards, tom lane