cryptic notes from Twice-Weekly Hack Hour, 2014-03-24

Zooko Wilcox-OHearn zooko at leastauthority.com
Mon Mar 24 22:58:35 UTC 2014


.. -*- coding: utf-8-with-signature-unix; fill-column: 73; -*-
.. -*- indent-tabs-mode: nil -*-

LAFS Twice-Weekly Hack Hour, 2014-03-24
=========================================

in attendance (or at least innocently bystanding on IRC): Zooko (scribe),
str4d, dreid, Riastradh, daira (LAST MINUTE, TIMEBOX-BUSTING, MEETING
PARATROOPER)

not in attendance although we wished he were: Brian, author and maintainer of
foolscap, author of half of Tahoe-LAFS, and much else besides

Topic: How to make foolscap work best over Tor/I2P/IPv6/etc.

str4d thinks the next step is to make foolscap (optionally?) use NaCl instead
of SSL for encryption. He's opening a ticket on the foolscap trac about that.

http://foolscap.lothar.com/trac/ticket/219

dreid pointed out that the Python libraries that implement NaCl also have a
cffi dependency, like his (dreid's) "cryptography.io" library that has
recently been the subject of Zooko's angst because of
https://tahoe-lafs.org/trac/tahoe-lafs/ticket/2193#comment:6

Zooko replied that well, that reminds him of how much he would like instead
upgrading foolscap, to eliminate the dependency on foolscap in favor of a
custom protocol embedded into HTTP:
https://tahoe-lafs.org/trac/tahoe-lafs/ticket/510

dreid said everyone should use endpoints for everything.

Zooko replied that in a #510 world, Tahoe-LAFS would be delegating that to
Twisted, using whatever the state of the art is for HTTP client and HTTP
server in Twisted.

dreid suggest that it might almost be time to consider jumping straight
to HTTP/2. Zooko had never thought of that before, and said maybe that
*would* be a good idea. Two of the best things about switching from
Foolscap to HTTP are that it could encourage implementations of LAFS in
Javascript in-browser [*] and that it could encourage implementation of
LAFS in different programming languages. Possibly jumping straight to
HTTP/2 could achieve both of those, while also offering performance and
perhaps sanity benefits compared to HTTP/1.

str4d said he had no qualms with it — I2P will take whatever protocol you
can throw at it. :P

Next meeting will be Monday the 31st (unless someone other than Zooko
organizes one for this Thursday the 27th). Aaron Cordova wants to show us
a new frontend something something written in Javascript. Aaron is
awesome because he invented Accumulo and
https://docs.google.com/file/d/0B9iCsXQ_HuEpN2NlNDgwMTMtNWRjOC00YzMwLWExYjMtYzVmM2FiZGRhNjA4/edit?pli=1&hl=en
so Zooko is interested!

Daira arrived at the end of the first hour, and the meeting was
overflowing into the second hour when Zooko (scribe) had to leave to do
chores in preparation for flying to the Princeton Cryptocurrency Workshop
tomorrow, so these notes end here. ☺

[*] Yes, I know that makes no sense until someone (else?) solves a
    different unsolvable problem. We try to solve one unsolvable problem
    at a time!


-- 
Regards,

Zooko Wilcox-O'Hearn

Founder, CEO, and Customer Support Rep
https://LeastAuthority.com
Freedom matters.



More information about the tahoe-dev mailing list