[tahoe-dev] Weekly Dev notes, 20-Dec-2012

Brian warner at lothar.com
Thu Dec 20 17:36:20 UTC 2012


It's our last weekly chat of the year! Here are some notes:

in attendance: Zooko (scribe), Marlowe, freddyb, amiller, David-Sarah,
Warner (late)

* translation: Rana from the Tor project walked Marlowe through
     workflow; Their git repo is checked by transifex. Transifex git
     pulls their repo. The one thing we have to do manually is that
     transifex has no way to check for malicious coding added in
     translations, like XSS/injection-style attacks. Transifex maintains
     a glossary of "one-offs", words that appear. voulnet does
     translations for Guardian Project and Tor and would be interested in
     doing Tahoe-LAFS as well.

* Tenerife, Blake2

* ticket notes:
   #937, #1159, #1539: land them!
   #1240: probably ready to land
   #1525: if tests can be made to pass, land it!

   #1298: trivial, let's land it. If DS doesn't land by this weekend,
          warner will.
   #1643: DS will poke at it, else warner will try to fix
   #1679: warner will try to write a test, test_client.NodeMaker
   #166: DS will consider 166-3.diff, if no objections by this weekend,
         warner will land. The -d issue can be figured out for 1.11
   #1732: warner will try to write a patch this weekend
   #1802: warner will write the patch
   #1735: warner will review and land
   #1777: warner will land
   rest of 1.10 tickets: can be landed if simple, else kick out to 1.11

* 1.10: warner will make a push this weekend and over holidays. Goal is
         to get a 1.10-alpha1 out by xmas, then aim for 1.10-final by
         late January.

* leasedb
   * warner still needs to review davidsarah/1818-leasedb
   * we still need to implement starter-lease code before landing
     1818-leasedb, probably in january
   * DS will rebase 1819-cloud-merge on top of 1818-leasedb
   * DS decided to stick with separate SI,shnum columns instead of a
     single merged shareid column: sqlite accepts foreign keys and unique
     keys and handles primary keys well enough that quota-measuring
     queries (finding unique lists of shares owned by a given account)
     are simple, or at least they can't be simplified enough to justify
     the denormalization hit
   * DS will consider adding the additional leasedb state-transitions to
     the docs (COMING->GONE when the upload process is interrupted before
     the backend creates any part of the share object, STABLE->GONE when
     the shares are deleted out-of-band, GOING->GONE when the share
     deletion process is interrupted but completes anyways), even before
     those transitions are actually implemented.

There will be no dev call next week on 27-Dec-2012. We'll meet again
03-Jan-2012.

happy holidays!
  -Brian



More information about the tahoe-dev mailing list