"Markus Wanner" <markus(at)bluegap(dot)ch> wrote:
> Another line of thought: isn't this like READ COMMITTED for just
> the first operation in a SERIALIZABLE transaction?
I've mulled it over and I have two different logical proofs that
this is safe; if anyone is dubious I'd be happy to share.
This seems likely to be of significant benefit in some workloads,
and I can't see anywhere that it is likely to cost much. Any
objections to adding this to the TODO list as a performance item?
-Kevin