db-ojb-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Martin Kalén <mka...@apache.org>
Subject Re: [OJB 1.x] Separation of PB-api from the kernel
Date Thu, 14 Apr 2005 12:48:16 GMT
Armin Waibel wrote:
> For 1.x I suggest to completely separate the PB-api from the kernel. So 
> the PB-api will be an API like ODMG, JDO,... a kind of top-level api.
> Further I want to avoid that OJB.java will became a conglomeration of 
> user and internal used methods. This class should be as simple as 
> possible, because it will be the first class of OJB a newbie will use.
> Then the top-level api using the PBInternal (PBI) instances in the same 
> way as jdbc-connection instances and the PB-api is a thin layer above 
> the OJB kernel.
> The PB-api separation allows to extend the PB-api with additional 
> features or to replace it with a more sophisticated layer with 
> UnitOfWork/dirty detection stuff.
> Any other suggestions or comments?

This looks great to me! I think this is the natural progression now that
so much of the old "PB-API" (=new kernel) is being reused from different

A clean separation of kernel functions from PB API makes, just as you say,
PB API qualify to the same league as ODMG et al. That might be less
confusing for new users, and is certainly cleaner in the code and promotes
the idea or re-using kernel functions from all layers (using PB API from
ODMG might seem a bit "twisted", but using a kernel from ODMG is natural).

It's less code to maintain and avoids that the same problems are being
solved in different ways in the different API:s.

On a similar note: what happened to Brian's Grafolia? Was that not
also a way of keeping object graph "excercises" centralized in code?
(Did not dig deep enought at the time to really know.)


To unsubscribe, e-mail: ojb-dev-unsubscribe@db.apache.org
For additional commands, e-mail: ojb-dev-help@db.apache.org

View raw message