On Apr 12, 2005 7:16 PM, Martin Taal <[EMAIL PROTECTED]> wrote:
> Hi Thomas,
> I have tried my luck on the collection-factory thing and have made
> changes to ojb to allow a collection-factory in the collection
> descriptor. I have tested with my own test cases and have run the ojb
> test suite (which gave the same results as before my changes).
> 
> I have added the use of the collection-factory in the same locations
> where the collection-class was used.
> 
> However, what I noticed is that the actual creation of the collection is
> a bit spread out, the PersistenceBroker, MtoNCollectionPrefetcher and
> CollectionPrefetcher each have their own createCollection method. Which
> basically do the same thing. Also the PersistanceBroker forwards the
> createCollection to a componentcontainer.
> In different locations (in the above classes) a check is done if a field
> type is an array in which case the collection creation is also handled
> differently
> 
> To stay in line with the current approach I have added the
> collection-factory in all of the above locations.
> 
> But would it be an idea to put the create collection code in a more
> central location (CollectionDescriptor maybe?) and/or use a default
> CollectionFactoryImpl to also cover the current createCollection approach?
> 
> I also noticed another discussion between you and others about an issue
> with the collection creation (jira, OJB-19).
> 
> I am happy to send in my changes (as a patch) for the collection-factory
> but as you can see I am not sure if the changes should may be go a bit
> further.

Could you add it to the OJB-19 or, better yet, create a new issue ?
I'll work on it then.

regards,
Tom

---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]

Reply via email to