| Version 60 (modified by zooko, at 2013-02-28T17:52:46Z) (diff) |
|---|
Weekly Meeting
There is a weekly conference call using Google Hangout. The agenda is set prior to each meeting and is focused on topics directly relevant to Tahoe-LAFS development. The connection URL is posted on this page and IRC a few minutes before the meeting.
| Who | Core developers or interested community members |
|---|---|
| What | Dev Topics Meeting |
| When | Thursday 17:00Z (9:00am Pacific) |
| Where | Google Hangout, IRC fallback (see below) |
| Why | Voice/video interaction complements IRC/mailing list |
| How | See Agenda below |
(*) Z (Zulu) time refers to UTC; check where you are or see the countdown.
Etiquette
- It's okay to talk about things not on the pre-arranged agenda. (But it helps to announce in advance if you know there's something you'll want to talk about! It encourages other interested people to prepare and helps them avoid accidentally missing a meeting that they really would have wanted to attend.)
- Participants may have high latency; if you'd like to speak, say "turn" or raise your hand and wait for silence.
- Use the hangout chat feature for textual chat, not the IRC channel.
Agenda
URL: https://plus.google.com/hangouts/_/f7cd6b3f4d4fe39ab5fed24e60ad90ef71aaa55f?authuser=0&hl=en
Upcoming: 2013-02-28
This will be a "TESLA COILS AND CORPSES" meeting — it is about science, big new features, writing papers about our work, etc.
If you're more into engineering, debugging, documentation, making stable releases, etc. then stay tuned for a future meeting about "NUTS AND BOLTS".
Agenda: (attempt number 3) Iraklis's research in applying Ristenpart's Message-Locked Encryption to LAFS. This requires extending the model of Message-Locked Encryption, and it suggests interesting directions for future extensions of LAFS.
To prepare for this meeting:
- Read the paper about Message-Locked Encryption and compare to the web page about LAFS's convergent encryption. See also my recent letter about ways that the "Message-Locked Encryption" paper doesn't model LAFS".
Proposed Future Topics
- Rainhill design review which David-Sarah will distribute; call for outside crypto expert reviewers ("TESLA COILS AND CORPSES")
2012-03-07
- ("NUTS AND BOLTS"): update on Tahoe-LAFS v1.10 release process
To prepare for this meeting:
- Look at 1.10.0, contribute patches, tests, docs, or comments.
Notes / Archives
2012-02-21
- ("TESLA COILS AND CORPSES"): (attempt number 2) Iraklis's research in applying Ristenpart's Message-Locked Encryption to LAFS. This requires extending the model of Message-Locked Encryption, and it suggests interesting directions for future extensions of LAFS.
2012-02-14
- ("TESLA COILS AND CORPSES"): Iraklis's research in applying Ristenpart's Message-Locked Encryption to LAFS. This requires extending the model of Message-Locked Encryption, and it suggests interesting directions for future extensions of LAFS.
2012-01-31
- ("NUTS AND BOLTS"): Tahoe-LAFS v1.10, new WUI, "timeline" visualization, https://LeastAuthority.com commercialization of Tahoe-LAFS
…
[some weekly hangouts were skipped due to travel]
2012-01-10
- ("NUTS AND BOLTS"): Tahoe-LAFS v1.10
2012-12-13
- ("NUTS AND BOLTS"): documentation and internationalization
- ("NUTS AND BOLTS"): Tahoe-LAFS v1.10
- ("TESLA COILS AND CORPSES"): new secure hash function
2012-12-06
- ("TESLA COILS AND CORPSES"): Proofs-of-Retrievability!
2012-11-29
- ("NUTS AND BOLTS"): documentation
- ("TESLA COILS AND CORPSES"): distributed introduction, defense against rollback attack
2012-11-22
- ("NUTS AND BOLTS"): work on Tahoe-LAFS v1.10 tickets!
2012-11-15
- ("TESLA COILS AND CORPSES"): async notifications
What a lot of people really want is an alternative to Dropbox — something that functions very like Dropbox but without exposing your plaintext to spying and corruption. David-Sarah implemented a part of this with the drop-upload feature. It seems to me that the blocker which prevents Tahoe-LAFS from doing the rest of it is that LAFS clients have no way to get an asynchronous notification that a file has changed (i.e., so that they don't have to poll to find out if the file has changed). So: could we add that? Why not just define a remote interface offered by LAFS clients to LAFS servers. The remote interface is "hey_you_this_file_has_changed(storageindex)".
2012-11-08
("NUTS AND BOLTS"):
- #1240; Is it done? (I think it still needs fixed tests.) Can we commit it to trunk and be done with it? Do we need to merge it with #1679?
- #1679; Let's write a test for it! Has The Dod had continuous good service since he applied the patch? Has nejucomo tried reproducing his bug and applying the patch?
2012-11-01
("TESLA COILS AND CORPSES"): Garbage collection: use cases and protocols;
- #1832 (support indefinite leases with garbage collection)
- #1833 (storage server deletes garbage shares itself instead of waiting for crawler to notice them)
- #1834 (stop using share crawler for anything except constructing a leasedb)
- #1835 (stop grovelling the whole storage backend looking for externally-added shares to add a lease to)
- #1836 (use leasedb (not crawler) to figure out how many shares you have and how many bytes)
- #1837 (remove the "override lease duration" feature)
And the associated mailing list thread.
2012-10-23
Topics: Ticket #1240, Tahoe-LAFS birthday party, future agendas.
Attendees: Zooko, David-Sarah, amiller, ambimorph, nejucomo.
Concise summary:
- #1240 debug session - Incorrect caching logic suspected.
- Tahoe-LAFS birthday party - Saturday, 2012-10-27 at 23:00z - Each locale connects with a projector + google hangout.
- Future agendas - next week, Proof of Retrievability paper review; two weeks out, Rainhill review with outside crypto reviewers.
Finely detailed notes in MeetingNotes_2012_10_23
2012-10-16
Topics: Proof-of-Retrievability
Attendees: Zooko, David-Sarah, nejucomo
From IRC, Zooko summarizes (edited for spelling):
- "The reason we can do better than the previous state of the art is that we've expanded the setting to multiple servers and multiple clients, which opens up new defenses for the good guys against Ponda Baba."
- "It is useful way to think, to start with the completely safe camouflaged download protocol: just run your normal verifier, but tell it to save what it gets, and then to talk about how to make it faster without breaking camouflage."
- "There are two dimensions of how you might be able to make it faster: the dimension of multiple downloader/verifiers and the dimension of multiple servers."
