Re: Oracle Bug?

From: hpuxrac <johnbhurley_at_sbcglobal.net>
Date: 23 Apr 2005 17:37:07 -0700
Message-ID: <1114303027.928215.102350@f14g2000cwb.googlegroups.com>

> I was assuming, obviously incorrectly, that ORA_ROWSCN could be
> used to determine (in a sense) which COMMIT was responsible for
> a specific row.
>
> My assumption was that TX1 and TX2 being part of a single transaction
> would have the same SCN. That the autonomous transaction, acting in a
> sense like another user logging on with a separate session, would
have
> its own SCN: Obviously not the case. But why still escapes me.

Someone else earlier replied with this:

---

I don't understand what part of it seems wrong to you. For each insert, the function returns the SCN in effect at that time the function was evaluated. Oracle can't give out the SCN that will be in effect when the TX commits, because Oracle won't know what that SCN is until the commit happens, at which time it can't retroactively change the
SCN that it reported previously.

---

Does this not explain the why to you?

What specific questions do you still have? Received on Sat Apr 23 2005 - 19:37:07 CDT