Instead of storing todo.userId, use publish composite to control th…

[tmeasday]

…e publication.

[stubailo]

We need to update the method tests, since they check that the denormalization worked.

[tmeasday]

Yup, was going to do that, just wanted to check with you that it still made sense given this comment: #48 (comment)

[stubailo]

Oh that's interesting. That actually might be a good reason to keep the denorm... we could probably get it to 0 lookups.

[tmeasday]

I'm not sure what the priority here is. There are a few competing things:

1. Achieving best performance (seems misguided? but we do it in other places)
2. Demonstrating the most patterns for the guide
3. Being a clear and simple example app

Actually all three things compete with each other. I'm not sure what's best. Let's discuss tomorrow.

[stubailo]

I think we should go with "4: Write what we would do in a real app" - and I think I'd probably do the publish-composite thing until it became a real performance issue. So I think we should take the PR.

[tmeasday]

This is ready to merge, I'll let you hit the button and double check @stubailo

[stubailo]

Looks great - I rebased on master