I forgot a couple of things:
* My code is on github at the moment: https://github.com/mvdan/tor_sample
This repo contains client.c, the code sample I used for my application, and server.c, what I am currently doing as part of the timeline.
* stats/stats.go in that same repo
This is a chunk of code I wrote last month to try and get approximate numbers on how would consensus diffs affect the users. Following proposal 140, we wanted to see for how long would they be useful, how much transfer would be saved and if it would make sense to keep nonrunning routers in consensuses for a number of hours.
I stopped writing stats.go because it was all approximated and not really of much use. I thought I'd continue it once the server side algorithm is done, at which point we can have exact numbers on how much the diffs will weigh. And then the stats will be of use to decide on the matters above.