A while ago TheServerSide.net published an article with Jeff Richter’s thoughts on the future of .NET assembly versioning.

If you read the article you’ll find that the Longhorn OS will seriously change the way that .NET itself is versioned. In fact, it turns out that to a serious degree the whole idea of installing “side-by-side” versions of .NET itself will go away when Longhorn shows up.

Oh sure, they have plans for a complex scheme by which assemblies can be categorized into different dependency levels. Some levels can be versioned more easily, while others can only be versioned with the OS itself.

What this really means is that .NET is losing its independence from the OS. In the end, we’ll only get new versions of .NET when we get new versions of the OS – and we all know how often that happens…

I’d say that this was inevitable, but frankly it was not. Java hasn’t fallen into this trap, and .NET doesn’t need to either. Not that it will be easy to avoid, but the end result of the current train of thought portrayed by Richter is devastating.

Fortunately there’s the mono project. As .NET becomes more brittle and stagnant due to its binding with Longhorn, we might find that mono on Windows becomes a very compelling story. mono will be able to innovate, change and adapt much faster than the .NET that inspired it. Better still, mono will remain unbound from the underlying OS (like .NET was originally) and thus will be able to run side-by-side in cases where .NET becomes crippled.

Hopefully Microsoft will realize what they are doing to themselves before all this comes to pass. Otherwise, I foresee a bright future for mono on Windows.