I'm happy to see that much reactivity of the KDEnlive community! Yay, we're alive !
First of all, I think putting the work J-B has already done in a branch. It's not a waste, just a pause ;-) It's easier to merge the branch "refactoring" (J-B were working in it before he did the merge) in the new master later on than porting bug fixes of a branch (that touch little parts of code every where) on a new architecture. And talking of that, not having new features on an open source software is not good for the community, it's doesn't attract developers for instance.
Laurent has done several commits, looking at the logs, I guess he knows the architecture and the work that would have to be done on the current master to fix it. I will myself work around the code of commit I've suggested in the first mail, and also on the current state of the master, to be sure that we are taking the good choice. As said Laurent in something like 2 or 3 weeks, we should be ready to make the move!
Thanks again for your fast answers!!