Annick,
This isn't an answer to your question, but it may be helpful...
Are you only dealing with individual records per transaction? At a
previous job, we had a large OR mapped VisualAge app, and found that
using a single sequence for all tables and prefetching enough key values
for all records to be inserted in all tables was the most efficient.
Store uses Integers and prefetches values for each table from a
different sequence.
If you are dealing with single records, using a sequence field has no
huge drawbacks, but you will need to requery to get the primary key into
the object in the image. Niall is looking to see Glorp does this
automatically. Postgres has an extension to INSERT which allows you to
get back the id of the last record inserted, however Glorp does not
currently support the extension.
If you have n-to-many relationships in your transaction, prefetching
sequences values allows you to populate the parent key value in the
child records (for a 1-n) or both values in a link table record (for a
many-many).
On 4/30/14, 1:46 AM, Annick Fron wrote:
> Hi,
>
> I am trying to use sequence in Glorp using Postgresql for an id attribute, but I get an error
> id_seq does not exist.
> If I look at the generated table, and the Smalltalk code, the code says type « serial » for column id should be replaced by « integer » which looks wrong to me.
>
> Any idea ?
>
> Annick
> _______________________________________________
> vwnc mailing list
>
[hidden email]
>
http://lists.cs.uiuc.edu/mailman/listinfo/vwnc_______________________________________________
vwnc mailing list
[hidden email]
http://lists.cs.uiuc.edu/mailman/listinfo/vwnc
Locked
