Early Adopter on SxS Deployment
"If the push was for SxS, then you would be free to make a lot more breaking changes to the Framework, as it wouldn't actually break apps.
"20meg is a small price to pay for apps that 'just work'."
[Early Adopter]
Not exactly. It would just wouldn't break apps till they were recompiled :-). I definately prefer keeping breaking changes to a minumum regardless of whether SxS is priority.
In any case, it doesn't really solve the problem, because how many developers are actually going to go line by line through each of the breaking changes and each of the lines of code in their app to make sure it doesn't break the app? In reality, any issues that appear will probably be resolved in the same way they would have if you had allowed the app to run on the next framework version anyway. Either (a) you make it a habit to test your apps and you will have issued a patch before the next framework version is being used in a "production" environment or (b) you don't test your apps, so a client is going to be the one to discover the app anyway. If anything, for those who fall into category b, allowing the app to run on the other framework might allow them to get those bugs out a little quicker :-).
20 megs may be a small price to pay when you are writing server software. Desktop software is another issue all together though, 20 megs is no small price to pay when your potential clients all have dial-up connections. In fact, with the clients I have worked with that have resisted .NET support for client apps, it has ALWAYS been the runtime size that ends up meaning the app cannot be written in .NET.