Tom Lane wrote:
> Neil Conway <nconway@klamath.dyndns.org> writes:
>
>>Now, is this a bug?
>>
>
> Good question. I don't think this is the only example of a
> non-self-consistent situation that could arise after a series of
> ALTER commands; I'm not sure that we can or should try to solve
> every one.
Ummm...at some point in time, PG will need to be able to dump and
recreate a database no matter what the history.
No matter whether or not "non-self-consistent situations" occur. PG
needs to be able to snapshot and restore current state, whether or not
it is a horror.
Or else you might as well state that, like MySQL, the only thing to do
is to knock down the database, tar files, and hope no one is interested
in 24x7 uptime.
When my clients ask about Oracle vs. PG I like to say "PG". They still
mostly say "Oracle" and I oblige.
--
Don Baccus
Portland, OR
http://donb.photo.net, http://birdnotes.net, http://openacs.org