"Simon Riggs" <simon@2ndquadrant.com> writes:
> COMMIT NOWAIT can co-exist with the normal form of COMMIT and does not
> threaten the consistency or robustness of other COMMIT modes. Read that
> again and think about it, before we go further, please.
I read that, and thought about it, and don't think I believe it. The
problem is that there are more interconnections between different
transactions than you're allowing for. In particular you need to
justify that the behavior is safe for non-transactional operations like
btree page splits and pg_clog buffer page writes. The idea that's
particularly bothering me at the moment is that after a system crash,
we might come back up in a state where a NOWAIT transaction appears
committed when its updates didn't all get to disk. "Database corrupt"
is a situation that threatens all your transactions...
> New commit mode is available by explicit command, or as a default
> setting that will be applied to all COMMITs, or both.
I dislike introducing new nonstandard syntax ("Oracle compatible" is not
standard). If we did this I'd vote for control via a GUC setting only;
I think that is more useful anyway, as an application can be made to run
with such a setting without invasive source code changes.
regards, tom lane