I recently
blogged about some issues we've had with Datadude, and this is a post about yet another problem. The problem arise when we use schema compare, making a compare between the physical database and the project. When we do this we get differences, that actually does not exists, and I believe, that this too is related to caching.
If you want to reproduce this behavior do this:
- Let two developers (on their own machines) ensure that they got the latest from source control and also verify that schema compare has no differences, when comparing the physical database and the project.
- Let Developer A do modifications to one of the stored procedure and then check in that changes.
- Developer A then does a Deploy of the project to the physical database.
- Developer B should now do a "Get Latest".
- When Developer B makes a schema compare between the physical database and the project the result will be a false difference on the changed stored procedure.
- If Developer B opens the stored procedure in the project into the editor, he/she will be able to verify that the latest version is present.
- Now do a refresh of the schema compare and there will not be any difference anymore, even though no change is done.
One of the most important features of a tool like schema compare is the fact that I as a developer can trust the result to 100%, otherwise the tool will be useless. Unfortunately false differences does not convey that feeling.