Tue, 30 Oct 2001 14:46:02 -0500
Seems like setting up a generic DataAdapter API is indeed the right
thing to do. We'll want specific adapters for PostgreSQL, BerkeleyDB,
etc... I haven't taken a look at the specific Microsoft implementation,
but is the idea here to recreate exactly the ADO.NET stuff? Or should we
take a look at how to do DB interactions in general?
This is an honest question: there are some things worth cloning, and
some things worth rethinking... but maybe the goal is pure cloning?
A Rafael D Teixeira wrote:
> ADO.NET has two subtrees, one uses a specialized DataAdapter for
> SqlServer, the other uses OLEDB, and can be used with specific OLEDB
> provider or go thru ODBC to use any ODBC driver. I think our first
> implementation can be a OLEDB-bridge to Linux-ODBC drivers and after
> it many people can contribute managed OLEDB providers or specialized
> The SQLServer DataAdapter exists because:
> 1 - With less layers performance is a bit better
> 2 - As XML is the basis for ADO.NET dataset, this adapter can leverage
> the XML support that SQL Server 2000 provides. That leads to a lot
> more performance for hierarchical data (and believe me, most of my
> Business Objects are deeply hierarchical).
> Rafael Teixeira
> Brazilian Developer