On 21/09/2007, at 5:31 PM, Emmanuel Venisse wrote:
Rahul Thakur a écrit :
I have been wanting to pop this questions for sometime now.
For generation/maintenance of continuum-model:
a) Why are we using Modello? More specifically what does it buy us? I
know the model is expressed as xml and separate from Java sources,
but I don't see a benefit of having it in XML as opposed to
annotations in java sources.
modello file generate for use all model POJOs and the jpox metadata
file. With it, we can define old fields that doesn't exist in the
nex version and the migration tool use the model to do the db
conversion. I don't think we can all of that with annotations.
The DB conversion is less than ideal though. I think the tools that
OpenJPA already provide for doing such things are superior (from the
brief look I've had).
I'm thinking the model conversion is a bit of a "seemed like a good
idea at the time" thing that I did... I'd be happier with coding it
up by hand now and making the store simpler. In making the store
simpler, and carefully mapping to tables, I think we can avoid the
need for conversion anyway by trying to only make additions.
I'd be willing to experiment with alternatives - perhaps it would be
best to do this piece by piece though, rather than one big conversion
(as the impl. currently leaks out all over - we need to better
isolate the lazy loading and fetching of data through the actions
before safely extracting pieces).
Cheers,
Brett
--
Brett Porter - [EMAIL PROTECTED]
Blog: http://www.devzuz.org/blogs/bporter/
