We had the phone call about the multi-op implementation yesterday. Only Gustavo Niemeyer was fooled by my accidentally saying Friday 4/14. Marshall and Camille had me figured out from the beginning and were able to attend.
I went over the current state of my ZK-965 github branch. That situation is that I think I have all the wire protocols and client changes in place. I have a few simple tests that currently fail (of course). The missing bit is how to put the multi-op into the commit queue and how to do the actual operations to the database in an atomic fashion. Marshall is digging into the code this weekend to scope out what the next steps are. I volunteered Ben and Mahadev to answer questions (again) about how to dope out the commit data-paths. I also volunteered after the meeting to add anyone who is interested as a collaborator on the github repo for the branch. The only condition I would impose is that any code you produce must be authorized to go back to Apache. If necessary, I will reformat the changes into a patches submitted by different people and have those people submit the patches separately (but that is a pain).
