Managed Extensibility Framework Preview Available

re: Managed Extensibility Framework Preview Available

Robert Mühsig — Fri, 06 Jun 2008 08:06:36 GMT

This sounds like a "plugin-model". What is the difference between System.AddIn and MEF?

re: Managed Extensibility Framework Preview Available

BradA — Fri, 06 Jun 2008 08:54:52 GMT

>>This sounds like a "plugin-model". What is the difference between System.AddIn and MEF?

Robert -- someone already asked ;-) see a detailed answer here:

http://forums.msdn.microsoft.com/en-US/MEFramework/thread/cf6b7cbc-1123-4b32-9810-c235d9606b66

re: Managed Extensibility Framework Preview Available

Eisenberg — Fri, 06 Jun 2008 09:05:08 GMT

Here's another question. What's the difference between this and Unity or any other DI container? Furthermore, haven't we (the majority of the community that's been using DI for a while) established that using attributes for injection is not the best approach? Perhaps I'm missing a nuance of your implementation.

Managed Extensibility Framework Preview Available

DotNetKicks.com — Fri, 06 Jun 2008 09:45:06 GMT

You've been kicked (a good thing) - Trackback from DotNetKicks.com

re: Managed Extensibility Framework Preview Available

Shaun — Fri, 06 Jun 2008 14:31:52 GMT

I don't really see what this gives you over just using a DI container

re: Managed Extensibility Framework Preview Available

Robert Barth — Fri, 06 Jun 2008 19:40:55 GMT

In a broader sense, reuse is not really a technology problem, IMO, it is a problem of being able to discern the future. What hooks does one permit? Where? You can't really know until you've come in contact with new requirements. You can guess, but that's all you'd be doing. As long as any framework requires knowledge of the future, it wil fail to deliver the promise of reuse.

Not that I don't think this API isn't a good idea: it delivers a standard way of doing something that is often required in a software solution. However, alone, I'd say it won't increase reuse one iota. That is the work of smart software architects, who, on their best day, can only make educated guesses about the future.

re: Managed Extensibility Framework Preview Available

David M. Kean — Fri, 06 Jun 2008 21:25:41 GMT

Eisenberg: I've answered your second point here: http://davesbox.com/archive/2008/06/06/customer-question-does-mef-allow-dependency-injection-without-using-attributes.aspx. With regards to your first point, this, not surprisingly, keeps getting raised - I'll try and get someone to post a blog entry on this.

Robert: Thanks for the feedback. You are right, its very hard to predict the future but that's exactly why we are releasing an early look on what we are working on - so that customers can have an impact on where we take MEF. At the same time, we're also working with some large customers internally at Microsoft (including Visual Studio), to make sure that we not just taking a 'stab in the dark' at guessing what customers want.

The way in that it promotes reuse, is by encouraging loosely coupled architectures. Yes it will be still on the software architects shoulders to get this right, but having MEF in their toolbox, means that its one less thing they need to worry about and they can focus on designing a system that matches their customer's current and future needs.

re: Managed Extensibility Framework Preview Available

Steve Naidamast — Mon, 09 Jun 2008 15:57:20 GMT

Code Reuse is not a problem with technology but has been primarily determined as an organizational problem whereby it is not encouraged in the IT areas.

Code Reuse requires repositories at different levels but rarely are they ever part of any IT organization's infrastructure. The second major factor here is a lack of discipline that reuse requires as much as repositories.

Providing a new framework to promote code reuse won't do much given the current situations within companies, it will simply promote the use of another framework...