I'm doing the June presentation at
SDNUG on Real World Code Generation. The session basically covers the back-end system that has been put in place at the company I'm currently working with, and uses
.net Tiers,
VSTS DB Pro and
CodeSmith to fully generate the back-end data layer. I've always been more a DataSet than an ORM guy (
this was one of my many impassioned defences of the noble dataset), but .net Tiers has convinced me that ORMs are mature enough to compete and beat DataSets in terms of features.
The thing that makes the solution that we've developed presentation-worthy are:
We aren't reliant on dynamic SQL (though we can support it)
We have a strong story for extending the generated objects on the C# side (i.e. post-generation)
We have a strong story for extending the generated objects on the TSQL side (i.e. pre-generation)
Re-generation works very well - this one is critical
Binding to the UI works
Events from changed properties work
Everything is under source control, and no manual merging is required by developers
We have a story for detecting optimistic lock problems
We can add auditing to any table in the database
We can get DataSets when required
Performance is good
For those not in Sydney who want to see the presentation, contact your local INETA user group, and INETA APAC can pick up the travel expenses (assuming you're not too far from Sydney) to bring this presentation to you.
On the project I'm on at the moment, we're making extensive use of VSTS DB Pro as part of the process for generating the DAL and keeping it under source control. I just found
this post where
Gert has announced that there will be a MSBuild task for automating schema compare (which is an important step in our build process) as part of the DB Pro Power Tools. Can't wait for this release.