Question

So I’ve run into something and was wondering if you could help me out. I am very new (this is my first project in these languages) to MS MVC 2 and Entity Framework 4.0. I’ve drawn up a model using the Model First approach and the graphical designer. I didn’t have a SQL database for it yet and I figured it was just working disconnected, but maybe I just forget that I had specified a connection string somewhere. So I go about my business getting the MVC components up and running using the newly created types from EF. I am attempting to using the Repository pattern as well as Ninject for dependency injection. So I had a lot of time before I had to worry about actual data storage. Which brings me to my issue: I’m now at that point.

I thought it was attaching to a .mdf file and running that off a SQLExpress instance, but I couldn’t really find anything to support that. I didn’t have SSMS installed so I couldn’t figure out how to properly check. Eventually got that up and running and no db. So now I’m confused as to where this is actually wired, if at all. I decided that it’s too weird and want to start over a bit. I generated the SQL script for the Model and create a new db in SSMS to run the script against. Run the script and all is well. Then I go into VS and run a Database First scenario. So the Model now shows all my tables and effectively screws up a few fields by specifying them as keys when they shouldn’t have been. Fixed that up and now I’m back to figuring out the issue that frustrated me in the first place.

I have a User table that has a userId field that is supposed to be a Guid, but during the original modeling I missed it and it was a varchar field. I’ve not been able to fix the mapping underneath it to change the underlying ( not sure the actual terminology for it ) field to a Guid. I changed the Model’s type but nothing updated the database. This has basically put me in an odd position. I believe I’m supposed to allow EF to manage the SQL for me without my involvement, but if this is true how can I push a change like that. Am I restricted to dropping the entire database everytime I need to make a change to the schema?

Sorry if this is the noobest of noob questions, but I am trying to figure all this stuff out and could really use some input. Last language set I used was ASP.Net 2.0 with ADO.Net typed datasets. Am I better off switching to a Code First approach? If I were to do that do I have to start managing the SQL myself again?

I believe this is my first question here so, my apologies if I violated anything; I think I’m clear though.

Answer 1

Well – EF stores connection string in Web.config file. Your suspicion that EF push you to delete database and create a new one when using model first is partially correct. The good news is that whole this process is driven by a worklfow which can be modified or replaced. You don’t have to do it yourselves. You can download Entity Designer Database Generation Power Pack which contains several new worklfows and templates which allow incremental building of the database via model first approach. This has two disadvantages:

• You must have either VS 2010 Premium or Ultimate because it uses database features from these editions.
• Not every change to the model can be processed incrementally. Anything that can lead to losing data will end with exception. I expect that this would be the case when modifying string to Guid. The solution in such case is either modify DB manually or delete it and recreate it from scratch.

Code first will probably not help you because this approach is currently missing any database evolution but it has fully automated process of recreation the database if model changes.

Btw. you don’t need SSMS to check the database. SQL server also offers some tools (which must be separately installed) with command line utilities to access the database – osql or sqlcmd. You can also use Server Explorer in Visual Studio Professional and higher.