Re: statement-level atomicity

[Ann W. Harrison]

Jim Starkey wrote:
> "Repeatable read" and "read committed", definition, are not consistent 
> and not isolated, so, no, they aren't ACID.
> 

Jim's talking about the standard form of Repeatable Read which has
phantoms - newly committed records will be seen, though no records
that have been read will be deleted or updated.  The MVCC version of
repeatable read is repeatable.

Cheers,

Ann