I don't feel competent to contribute to the kind of test framework you're proposing, because I wouldn't be using it myself. I think that the right progression should be something like this:
1. You discover that you have a similar need in several tests and write some support code. 2. You refactor that into a small framework that solves the problem well for most of your current and some of your future use cases. 3. You clean it up and release it on pypy and solicit feedback. 4. You and others who use it iterate a few times. 5. Eventually you suggest adding it to the stdlib -- or perhaps to the tulip project (since I assume your framework might have dependencies on Twisted and/or Trollius, which isn't really appropriate for the stdlib). It seems you're currently in stage 2, but you seem to be asking to jump straight to stage 5? -- --Guido van Rossum (python.org/~guido)
