Back in 2003, Microsoft launched a project that they hoped would “capture people’s ideas, requirements and hopes for software” and turn them into distributed applications. It was variously described as the “Excel” of the database world; a people’s data repository. In short, this project, codenamed Oslo, set out to create a general-purpose software-modelling platform that would allow non-developers, i.e. Business Analysts and so on, to create applications from diagrams, using a simple visual tool, called Quadrant, and a ‘repository’. Included in the vision was a modelling language (initially called “D”, but later “M”) for “people who like textual programming” that could be used to create schemas and domain-specific data models.

Now, after many tortuous twists-and-turns, and years of diminishing goals, we hear that Quadrant has finally been abandoned, and M is to be ‘refocused’.

In retrospect, the vision seemed over-ambitious from the start. Anyone who has ever suffered the task of turning UML diagrams into a database schema would have quaked at the plans, initially expressed by the team, to “enable users to capture domain knowledge in domain-specific views”, and support “the development of BPMN (Business Process Modelling Notification) workflows and UML (Unified Modelling Language) services”. Slowly and painfully, the project got ‘refocused’ towards reality.

Eventually, Oslo found itself under the ownership of the Data Programmability team, responsible for Entity Framework, ADO.NET and so on. It was renamed “SQL Server Modelling” and was to be integrated, at some point, into the SQL Server platform. Many developers, whose primary interest in the project was “M”, despaired at the loss of their dream of a domain-specific language (DSL); it seemed clear to them that Oslo would gradually became just another data modelling and querying tool for SQL Server, so they promptly lost interest. There is little moving in the support forums but tumbleweed.

What, if anything, will emerge of the original vision? Does the loss of Quadrant mean that we’ve also lost hope of a reasonable domain-modelling tool for .NET? The rise of Entity Framework has meant that there is now even more of a need for a domain modelling tool, and the Eclipse Modelling Framework has shown that it can be done for Java. Surely, such a tool, which could work with Entity Framework and C#, would be welcomed by developers, and is well-enough scoped to be developed in a reasonable timescale?

Cheers,

Tony.