I don't at all disagree with the tone or message of your post, but there are valid reasons to switch containers.

For the same reasons we upgrade to new features in the .net fx or make architectural changes in our apps to make them more robust, or more succinct, or more intention-revealing, or more maintainable, or w/e, we may change a container to make it less xml-heavy for maintenance/upkeep reasons or take advantage of some new convention based config to make attributes go away and decouple a bit. Or maybe there's speed or size improvements to be had.

Just sayin it seems pretty knee-jerk to hate on container switchers.

Thank you for submitting this cool story - Trackback from progg.ru

I agree with Scott on this, the main difference between the different containers is how they are being setup. A choice for me is how readable the configuration is for others in the project.  Also some are more sophisticated in some scenario's that you may not have needed at the start and you went for what you were convertible with and later realized it would help you in development, so why no switch container.

I agree with you about it not being a feature that you use in your advertisements :) but it _can_ give an _initial_ impression on how decoupled the code is.

Two random thoughts:

Yes, switching to another container, because it's advertised as being more sexy is plain stupid. However there are much deeper differences between container than just registration/retrieval syntax. This is simple DI, and mature containers (StructureMap, Spring, Windsor) offer much more beyond simple DI. When you realize that, and want to switch to one of these containers to reap benefits of thsese additional features to simplify your architecture and/or make faster progress with your application, this is a valid reason.

It all depends on the context though. When you have invested in a good, mature container, adding features/modifications you need i's more often than not, just a matter of finding a right extension point and using it.

This rush towards lambda-driven containers remind me of similar movement in mocking frameworks space, where people started using MoQ when it was very new because it had such a cool name and used these new funky lambda magic, only to realize it lacks features, and to move back to old .NET 2.0 smelly Rhino.Mocks.

If something was good few months ago, is it worse now because it does not use latest super-duper kung-fu API/language?

You're absolutely right that a choice of container is only a matter of personal preference, and has very little to do with your application. In fact, that's really the only reason Ninject even exists -- I didn't like the existing offerings, so I wrote my own. :) If someone prefers another container, then by all means they should use it, and I don't think anyone should waste the time to switch between containers without a very specific reason in mind.

Tim Barcz posted a great article on how the specific choice of an IoC container doesn’t really mean much

@Nate

What exactly were you dissatisfied with? Why not join an existing project and improve the existing tool, instead of starting from scratch?

It would be helpful if we could define an IoC container to some core functionality ; GetInstance, CreateInstance, BuildUp.  The first two are easy to create adapters are easy to create adapters for. The third one, because it relies on attribute decorations, is difficult to create an adapter for.  Meaning that it requires code modification and recompilation.  It seems that as a community we need to define some standards for IoC containers.  Then it wouldn't matter which one you use.  You could easily swap one out for another.

Additionally containers are beginning to get more feature rich with features that apply Aspect-Oriented Programming concepts.  Making it more difficult to swap one container for another.

Are we making IoC containers more complex than they need to be?

I completely agree with Krzysztof Koźmic.

My main reason for sticking with Castle is more about the facilities that are supplied than the container.  IoC containers are dime a dozen, but it's the "extras" that make me stick.

However, I am no more loyal to Castle than I am to Nike, and trying out different IoC containers and the fearures they bring the to table is something I as a developer should keep on top of.  As transparent as an IoC container is, the easier it is to get what you want from it the better, Castle delivers this but as soon as something else delivers better value cheaper, I will switch :)

Thank you for submitting this cool story - Trackback from DotNetShoutout

I agree that an IOC tool isn't a feature, like a nail gun is not a feature of a house.

But I disagree that IOC tools are commodities and picking one over the other won't matter that much.  "Choosing one container over another won't make your application a success" - true, but choosing the wrong one can make your application a failure.

Some of these tools are hobbyist products. Now that is not a dig on a specific tool, although it will likely be interpreted that way.  But the facts are that some of the tools you list have been in production, for years and years, in big huge projects and some haven't.

Things happen.  Bugs surface.  Developers spend weeks debugging and patching trunks.  Hobbyists end support and their products are obsoleted earlier than is optimal.  There is a public example of this in recent memory.

The point is: it DOES matter which container you choose.

In this post, I will share some of the best practices/guideline in developing ASP.NET MVC applications

It is better that we have a wrapper like JQuery, IoCQuery perhaps?

Thanks for sharing! The reason for your blog is always to document your life and give your personal thoughts

3Bcs0h Im thankful for the blog article. Fantastic.

Lm89uF Awesome blog article.Really looking forward to read more. Will read on...

Hello! akkdbck interesting akkdbck site! I'm really like it! Very, very akkdbck good!

CjJYXw Thanks-a-mundo for the blog.Really looking forward to read more. Want more.

ddgJy3 Major thanks for the blog article.Really looking forward to read more.

ZdTXcx Really informative blog. Much obliged.

Thanks so much for the blog post.Really looking forward to read more. Awesome.

wYC7Xb Say, you got a nice post.Really looking forward to read more. Much obliged.

sPqM38 Great, thanks for sharing this article.Really looking forward to read more. Keep writing.

mgLs4q Fantastic blog.Really looking forward to read more. Keep writing.