On Fri, Mar 11, 2016 at 6:31 PM, Thomas Munro
<thomas.munro@enterprisedb.com> wrote:
> I'm not sure what to make of the pre-existing comment about following
> HOT-chains and concurrent index builds (which I moved). Does it mean
> there is some way that CREATE INDEX CONCURRENTLY could cause us to
> consider the wrong tuple and miss an SSI conflict?
No, because the check is done entirely on the basis of the the index
page. The question may be arise if we discover that we also need a
conflict-out check here though, because it would be based on the tuple
that has been found by heap_hot_search_buffer.
--
Thomas Munro
http://www.enterprisedb.com