Re: Risk in using sequence number?

From: Keith Boulton <kboulton_at_ntlworld.com>
Date: Thu, 21 Feb 2002 23:03:08 -0000
Message-ID: <e2fd8.4996$H43.594296@news11-gui.server.ntli.net>

Sybrand Bakker <postbus_at_sybrandb.demon.nl> wrote in message news:pnqa7uc1th9gp1ipq4c445br6n0rp29hr3_at_4ax.com...
> Operators, not being trained monkeys, usually don't insert more than
> one record per second.

No information is given about what is generating the event and "usually" is not good enough.

> And if you really need the time you can always
> use dbms_utility.get_time, which is in hundreths of a second.
See above.

> Other than that, if this are non-human inserts, you should ask what
> you need the sequence for. Looks like an issue of lousy coding,
You can't reasonably say this without more information - take for example the situation where a remotely controlled substation is reporting outage. It can easily happen that a remote monitoring unit briefly records an outage, immediately followed by a resumption of service. Get the order wrong and you are reporting a device as out of use when it's working fine. You absolutely need to know the sequence of notifications.

> definitely non-relational.

does not compute - this has no relationship to relationality (if you'll pardon the pun). Received on Thu Feb 21 2002 - 17:03:08 CST