Could you state clearly why you are writing alternatives to MEF, ParallelFx, and XML serialization? Not that it matters for you, since self-edification is a worthwhile goal, but for others to be interested, you need incentive.
Ok, i'm willing to be sequential.
1st point: Full source code access.
2st point: Framework size.
3st point: It challenging to do the better product as do the guys at Microsoft.
In your opinion, why Linus Torvalds stated his own project formerly known as Linux Project, even he knows that Unix/Windows/QNX etc. OS-es already exists?
If you are asking me, why I'n doing so. I have the answer.
I'm tired, because nobody hires me, nobody supports me, nobody follows me. Do you really think that big fat cats in Microsoft will ever give me a chance to do the best i can do? No. Misrosoft is a suxx as a work place, because people like me are eaten for the breakfast by the corporate approach, so I'm smart enouth mice to understand it. Too many guys tries to be a one-man-at-the-top. Probably, I'm going to continue to develop software while working as system administrator in no-one-knows-it-company for the end of my life. On the other hand I can do better than Microsoft Teams and I will.
So, if you can't challenge the best of Microsoft teams for your own, give me chance to do this for you.
If you are from one ot the teams you mention it earlier (ParallelFx, MEF, etc), just be ready to know what i'm thinking about you.
I'm trying to prove that you can create by YOUR own code which can do the parallelization in several assemblies with 4-8K size, and it will work BETTER that in .NET FW 4.0 or ParallelFx. YOU can do better. YOU can make use of it. YOU. Just learn what is incapsulation, interfaces, inheritance, factory pattern, etc, exploring my code. It is a prove that YOU can do better.