Re: [Metware-devel] foreign key constraints break MetBeans unit tests

[Egon W. <ego...@gm...>]

Hi Steffen,

sorry, missed this email... almost back to zero in my inbox, and still
discovering new emails :)

On Fri, Jun 12, 2009 at 10:35 PM, Steffen Neumann<sne...@ip...> wrote:
> we currently have the problem
> that the recently introduced foreign key constraints
> cause SQL exceptions if you try to save a Bean which
> as no / a non-existing reference:

Yes, I noted this too by now, with that in memory database...

> org.postgresql.util.PSQLException:
>        ERROR: insert or update on table "metchar_putative_metabolites_identities"
>        violates foreign key constraint "metchar_putative_metabolites_identitie_metobserv_output_id_fkey"
>  Detail: Key (metobserv_output_id)=(67) is not present in table "metobserv_output".

Yeah, the 'problem' is that explicit foreign keys requires all beans
to be constructed... that is, you cannot do a partial construct of a
data structure, and have to give beans for all links you make...
actually, this would not be so bad...

> I currently see two alternative solutions:
>
> 1) Revert the foreign key constraints, such that
>   (possibly inconsistent) data can enter the DB,
>   hoping that some following INSERTs fix that.

Yes, that's sort of the current behavior...

> 2) Enforce the correct ordering of Beans,
>   and either catch the SQLException,
>   or verify via our exists() that the referenced entity is there.

This equals fixing the unit tests? What do you mean with 'correct
ordering of Beans' ?

Egon

-- 
Post-doc @ Uppsala University
http://chem-bla-ics.blogspot.com/