Re: 7.3.4 on Linux: UPDATE .. foo=foo+1 degrades massivly

On 26/04/2004, Guy Fraser <guy@incentre.net> wrote To Philipp Buehler:
> What if your not the only person accessing the database and someone else
> has an active transaction that was initiated before your transaction was
> committed?

There are row locks for a reason, too?

OTOH that's somewhat like handling a filesystem. if you have
a refcount=x. Just decrease the refcount for completed
transaction and after it this count drops to 0 the tuple can be purged.

> supported, but hopefully it will be configurable to allow 'time travel'
> when required, and allow for a reasonable time to be able to roll back
> transactions.

Hm well. a comitted transaction is one, right? I dont get your
time travel here now.. hmm.

Ciao
--
Philipp Buehler, aka fips | <double-p>

cvs -d /dev/myself commit -m "it's my life" dont/you/forget