Oslo – Not just for Norwegians any more

I’ve been looking heavily into Oslo, the new technology announced at PDC 2008. So what exactly is Oslo? Well I couldn’t find a simple explanation, so after digging into it all weekend let me see if I can take a stab at it.

If you are familiar with Sharepoint, you know that it provides you a bunch of web templates. You can take these and create certain types of lists. Documents, lists, forums, etc. What many don’t realize is that all of this gets stored in a “repository” that is in SQL Server.

Oslo takes this concept to the next level. It allows you to create your own “lists” if you will, of fairly complex data types. These are stored in a repository in SQL Server. Along with your data is a lot of meta-data. Oslo also provides a query tool to easily get data back out of the repository, along with runtime components you can use with your favorite programming language. Or, because it’s all in SQL Server you can bypass Oslo runtimes and go directly into the SQL Server repository using traditional tools like ADO.NET or Linq To SQL.

So how does Oslo accomplish this? By providing several tools to us: M the programming language; Quadrant, the graphical tool; and the Repository itself. Lets take a brief look at each one.

M is a new programming language that has three components: MSchema, MGrammer, and MGraph. MSchema is used to define a new chunk of data, it is a representation of how you want the data stored. The product of an MSchema definition is directly translated into T-SQL as a Create Table statement and stored in the Repository.

MGrammer is used to create a translation between one layout of information and the schema created with MSchema. Let’s say you had created an MSchema definition for album names, artists, and ratings. Then let’s say you had an input file that looked something like:

The Thirteenth Hour by Midnight Syndicate rates 5 stars.

Greatest Tuba Hits of 1973 by The Tuba Dudes rates 1 star.

You could create a language template in MGrammer that looked for the words “by” and “rates” and divides up the input into the appropriate fields in your schema. Then run the input file through the MGrammer layout and you’ve now got all that data into a format known as MGraph.

MGraph is a tree like structure that represents the transformed data. If I understand it correctly, you take your data, run it through the DSL you setup with MGrammer and it produces an MGraph. This MGraph can then be loaded into a database schema created with MSchema, passed off to a calling routine, and more.

Quadrant is the tool used to look at data once it’s in the Repository. You can browse data, and create different representations of the data in a tool similar to what you see with Office. For example, you can render a table created by MSchema as a tree, as a grid, as a list, or even as a graph. You can use it to show relationships between MSchemas, and write queries with it. Quadrant could be used by developers or advanced users to create a template representation of the data that could be given to other users to do their data analysis.

Quadrant is also highly extensible and customizable. You can write your own modules to add to it. Although to do so you have to write them in Python, which I have to admit leaves me scratching my head. I don’t have anything against Python, but I would have to imagine most developers who work with Microsoft tools are much more familiar with VB.Net or C#. I have to wonder why they picked a language most Microsoft developers are unfamiliar with and would have to learn in order to extend the Quadrant tool.

The final piece of the puzzle I have mentioned several times, it’s the Repository. The Repository is a database that holds everything about your schemas and data. Currently Oslo only supports SQL Server as the database for a Repository. Interesting thing though, Microsoft will be distributing Oslo under the OSP (Open Specifications Promise). This means a third party vendor could develop a back-end Repository engine so that an Oslo Repository could be stored in something like MySQL or Oracle.

Finally I will mention Oslo will be callable from your favorite .Net language, indeed the Runtime components as they are called are a critical piece of Oslo. There are .Net APIs which can be used to get and retrieve data from the Repository.

Microsoft is serious about Oslo. In a Channel 9 interview about M, I believe it was Chris Anderson who said there were 180 folks working on the Oslo team. Even though it’s early in its development, I get the strong impression Oslo will be a key factor in future of Microsoft development technologies, which is why I intend to invest time now to get up and running with it.

For more information about Oslo, and to download the current Oslo SDK CTP, see the site at http://msdn.microsoft.com/oslo .