[tor-project] Network team meeting notes: 28 Aug 2017

[Nick Mathewson]

Hi, folks!

Another week, another meeting.

You can see the IRC logs from
http://meetbot.debian.net/tor-dev/2017/tor-dev.2017-08-28-17.01.html .

Here are the notes from our pad:

=====
Network team meeting, 28 August 2017 (1700 UTC)

Announcements and requests:
    - PA request from HS team: We need some help with review-group-22
since all our
      prop224 tickets are blocked by this review group. We've already
reviewed most
      of the tickets; there are 2 left which would be great if were
handled by other
      team members since we have tons of other prop224 stuff to tend
to. Thanks! :)
    - 031 stable release? Triage, Bug fixing, ... so we can meet the
release deadline?

People out:
    - ahf (returning wednesday).

teor:
    - Last week:
        * Caught up on emails
        * Caught up on other admin
        * Re-planed PrivCount Experimental for the next few months
        * Tried to work out how to generate random floating point
numbers that are good enough for
          Laplace noise (current hidden service stats) and Gaussian
noise (PrivCount in Tor).
          Turns out floating point is really lumpy.
            https://trac.torproject.org/projects/tor/ticket/23061
        * Told people that I want to limit the amount of work I do on
fallback directories:
            https://lists.torproject.org/pipermail/tor-project/2017-August/001405.html
    - This week:
        * Work out how many onion addresses an HSDir should expect to see
            https://github.com/privcount/privcount/issues/397
        * Try to stop adding features to PrivCount 1.1.0 so we can
actually do the release


komlo (offline)

- Last week:

- Fixed up Rust protover from isis's review

- This week:

- Pair with isis to finish up Rust protover, submit it for wider
review hopefully by next week

- Send out Rust kata for the Rust hack day in Montreal, if people want
to work on it beforehand


nickm (out from Wednesday through next Monday)
   - Last week
     * Lots of review and merge, especially hidden service client-side stuff
     * Working on the aftermath there
     * Lots of tiny tickets in 0.3.2.x.
     * Starting to triage some 0.3.2.x stuff; discussed stragegy with isa
   - This week
     * Only around 2 days.  Planning to review patches and triage 0.3.2.x.
     * Didn't get 0.3.1.x-rc out last week. Maybe can make progress
this week? Still bugs to fix. #22752 seems blocker-ish.
     * Going to hack some low-hanging sponsor8 efficiency stuff, time
permitting.
     * going to try to empty my inbox
   - Queries

dgoulet:
    - Last week:
        * prop224 client upstream merge (YAY!!!) which lead to bug hunting lots
          of issues. Some relevant tickets I worked on are #23300 and #23308.
        * Reviewed tickets in review-group-22 including the big KIST one from
          pastly in #12541.
        * I've addressed some tickets in 031 milestone as well. More work there
          remaining for the stable release.
    - This week:
        * Wrapping up many bugs for prop224 so we can get it in a good state
          before the 032 feature freeze. This means continue integration
          testing and hopefully tackling #23310 sooner rather than later.
        * Follow up with pastly on KIST scheduler and try to make it in time
          for the feature freeze deadline.
        * I'll prioritize 031 tickets review/revision this week so we can meat
          our release deadline.

pastly:
    last week: NRL stuff
    this week: NRL stuff, meet with dgoulet about KIST

haxxpop:
    few weeks ago: run the hs desc fuzzing and found a bug #23233
[Good stuff! --dgoulet] << d(^ ^)
    this week: I will continue the fuzzing on the inner layer of descriptor

asn
  Last week:
   - We got the client-side of prop224 merged (#17242). We are basically 3/4 of
     the way done for prop224. The rest 1/4 of work is testing, bugfixes,
     unittests, and quality of implementation improvements! There are
lots of them.
     See our pad for more details:
https://pad.riseup.net/p/LPVY6AL_prop224_upstreaming
   - Spent lots of time testing our prop224 branches for bugs related to hsdir
     indices and overlap period. Found a few and fixed them: #23157, #23110,
     #23309, #23335
   - Wrote patches for more prop224 issues: #23056, #23343
   - Reviewed #19418, #22779, #22802, #17242 as part of review-group-22.
  This week:
   - This week we hope to get all the current prop224 blocker bugs merged
     upstream, so that we have a common branch to work from. See our pad above
     for more details.
   - We need to keep testing prop224 and writing unittests. We keep finding
     new bugs :-)
   - Maybe write a few more stuff in #23126 about how to do hsv3 statistics.



isabela

Last week:

- caught up on emails after looong vacation

- chated with networkers about 0.3.2 triage and all the long backlog
we have, how to organize it over the different upcoming releases (033
and 034)

- tried to coordinate dependencies TB team has on network team for Tor
Launcher UI udpate (progress bar and moat api)

This week:

- write sponsor4 deliverable completion report

- following up on dependencies (question for catalyst and isis and
nick?) - just checking in if things are moving on and if there is any
udpates/blocker we should be scary about (TB dependencies)

- plan to check in with sponsorR deliverables completion - and to
check in with sponsor8 work (how is it going etc, I need to pair this
up with what we wrote at M&E which is a looong story but no worries
#wegotthis)

- if all goes well I also plan on getting us started with roadmap
planning (for montreal)


catalyst:
    last week (2017-W34):
        - looked at floating point nastiness (#23061) -- I should
probably drop it but I want to make sure we explicitly say things like
"we don't care about holes in the range where we lose precision"
        - poked some more at bootstrap progress reporting
        - somewhat distracted by general awfulness in the world (and
family in Houston)
    this week (2017-W35):
        - travel arrangements for Montreal
        - figure out how to incrementally retrofit a higher-level
state machine that tracks bootstrap progress based on lower level
events (so lower level code won't have to break abstractions to report
boostrap progress)
        - review #23149
        - ask pointed questions about how much precision we need in #23061

isis:
    last 2 weeks:
        - vacation
        - brief amounts of work to fix some issues with the bridge
authority and talk to the intern
    this week:
        - moat backend #22817
        - montréal arrangements


-- 
Nick