Re: Import Question

On Sep 27, 12:59 pm, Hasta <hasta..._at_hotmail.com> wrote:
> In article <fdgqv9$jn..._at_news5.zwoll1.ov.home.nl>,
> frank.van.bor..._at_gmail.com says...
>
>
>
>
>
> > -----BEGIN PGP SIGNED MESSAGE-----
> > Hash: SHA1
>
> > Hasta wrote:
>
> > >> It's not just about constraints or the order of tables.
> > >> The original question was more generic, and yes, sometimes
> > >> you'll get errors because dependent objects do not
> > >> (yet) exist: procedures, functions, packages and views to name
> > >> a few, can either be invalid and need to be recompiled, or are
> > >> not created at all, which can happen with views.
>
> > > Woaw ! When is a view not created at all, Frank ?
>
> > Whenever you have a 'Synonym translation no longer valid'
> > (ORA-00940?) error. Howard will probably argue this is not
> > an import error, but a user error.
> > It is not - it is simply the sequence in which import
> > handles objects. Rerun the import with ignore=y
> > and rows=n, and it should be OK.
>

There was also a trigger bug in the 9iR2 timeframe (see Note: 270570.1). I was never convinced the bugfix docs were exactly correct, I wound up finding differences in 9206, IIRC between a schema export and one using QUERY= on the export, when changing usernames for the import. But I could be misremembering the details, I got very confused when I ran into it and only had time to workaround, rather than be educated by support. Hopefully this won't bother the OP going from 8 to 10.

But it highlights the necessity to try everything first, and be especially watchful about what triggers will do after import (also see 9208 patch notes).

jg

--
@home.com is bogus.
http://ap.google.com/article/ALeqM5jzS7pLAIUAhQxfgv4-Z5ngpXkRWg