[tor-reports] George's status report: July 2016

[George Kadianakis]

# Activities of July 2016

- The code for proposal 250 got merged upstream!!! This is the shared
  randomness subsystem that we've been developing for the past year, and marks
  the first deployed part of Next Generation Hidden Services (#16943)!

  moria1 is also running the code and voting for shared randomness. Through
  deployment we also found a few bugs that we fixed (#19551, #19555, #19561).

- Did multiple rounds of review on Nick's latest guard proposal (prop271):
   https://gitlab.com/asn/torspec/merge_requests/4
   https://gitlab.com/asn/torspec/merge_requests/3
   https://lists.torproject.org/pipermail/tor-dev/2016-July/011188.html

  We feel that the proposal is at a reasonable state for implementation now,
  and I prepared an implementation plan for the feature:
   https://lists.torproject.org/pipermail/tor-dev/2016-July/011234.html

- Discussed some more possible design improvements for the HS protocol when
  used in unstable networks (i.e. mobile phones):
   https://lists.torproject.org/pipermail/tor-dev/2016-July/011170.html

  Also wrote a patch for a reachability/performance improvement for hidden
  services when hosted on mobile phones. The patch concept seems to work
  alright in testing, but it needs some more code to handle some edge cases. If
  anyone is interested, here are the details:
   https://trac.torproject.org/projects/tor/ticket/19522#comment:6

- I talked with various people who were worrying that the onion addresses of
  next gen hidden services will be too big to be useful. Current onion
  addresses are 16 characters, and next gen onion addresses will be 52
  characters. I find this complaint legitimate so I started a thread on coming
  up with ways to maintain a small onion address size (e.g. 24 characters for
  128 bits of security):
   https://lists.torproject.org/pipermail/tor-dev/2016-July/011235.html
   https://lists.torproject.org/pipermail/tor-dev/2016-July/011240.html

  Unfortunately, both of my ideas were completely broken as demonstrated by
  Nick in minutes :( I still have some hopes that my second idea could work
  with some mods, and maybe I should find some time to look into it.

- Discussed the Tails server GSoC project with segfault:
   https://mailman.boum.org/pipermail/tails-ux/2016-July/001079.html