Scott,
There are at least two deviations in my opl changes
from Scott's design.
One is that I used a separate class, SqlInterface, for
generating sql strings. He has this, and the sql
process mechanism wrapped in the Relational Database
object.
Another is that I didn't provided for converting
fields to sql strings on a field by field basis.
getSqlSaveValue() in the AttributeMap class is the
right place for this. Instead, I provided asSqlValue
in SqlInterface. This doesn't allow formatting on a
field by field basis. Of course, we could have:
public String getSqlSaveValue(Object value)
{
if (format provided) use it;
else return sqlInterface.asSqlValue(value)
}
With my mostly Oracle experience, I haven't seen the
need for field by field. What's the real world say?
I put some quick and dirty stuff in the mapping
package.
I don't have good ideas for implementing some of this
stuff, but I feel it's better to progress to a working
model so others might exercise it or contribute ideas.
In any case, we need a working (not perfect) model
just to exercise the code for bugs.
btw, I have another idea which deviates from Scott's
design, but might simplify the implementation. The
idea is to use another mechanism for proxy maps.
Scott Ambler suggests using a subclass of
AttributeMap, ProxyAttributeMap, for providing the
proxy's attributes.
Instead, why not provide a mapping from class name to
proxy class name? So, for example, if you have a
"Customer" object, the map could specify a
"CustomerProxy" object as the proxy. Then, the
"Customer Proxy" has a set of AttributeMaps. I think
that this would simplify the backend processing, and
requires minimal work up front.
What do you think?
That's all for now, george
_________________________________________________________
Do You Yahoo!?
Get your free @yahoo.com address at http://mail.yahoo.com
------------------------------------------------------------
To subscribe: [EMAIL PROTECTED]
To unsubscribe: [EMAIL PROTECTED]
Problems?: [EMAIL PROTECTED]
