[tahoe-dev] Weekly Dev Chat notes from 2013-03-07

Zooko Wilcox-OHearn zooko at leastauthority.com
Mon Mar 11 06:27:05 UTC 2013

Sheesh, it takes a lot longer to write these notes up for the list
than it does to actually attend the dev chat.

in attendance: Oleksandr, Zooko (scribe), David-Sarah, Randall
"ClashTheBunny" Mason, Brian (late—one lash), Tony (late—one lash)

not in attendance: Andrew (ABSENT—two lashes)

sort-of-in-attendance (on the IRC channel): a lot of folks

Agenda: Tahoe-LAFS v1.10 (Nuts and Bolts)


This was awesome. We got lots of tickets closed, or at least pushed forward a
step or two. This was the most productive hour of ticket-closing work
ever. We're going to do it again next week! If you like doing code-review and
writing unit tests, you should join us.


You can see also use the Trac Timeline:


ClashTheBunny has some patches for IPv6 for Tahoe-LAFS and Foolscap (#867),
and has some tests of them but needs more thorough tests. However, we didn't
look at that this time, because we focused on tickets for Tahoe-LAFS v1.10.

Nejucomo said on IRC that he wasn't going to get around to manually verifying
that the bug we fixed in ticket #1679 was the same as the bug he reported. I
decided to just close the ticket for the following reasons: there *is* a bug
that we understand, there is a fix which is clear, there is a unit test that
exercises the bug that is red without the fix and green with the fix, and The
Dod manually verified that it fixed the problem in practice for him. The only
reason we've left the ticket open waiting for Nejucomo to test is in case
Nathan actually has a *different* bug than this one. But we don't need to
keep the ticket open for that eventuality.

ClashTheBunny did design review on #1732. He approved the design and made
good comments that showed that he actually had thought about it. Subsequently
Zooko added another design question and assigned it to Brian for design

We looked at #1926. It is a blocker because it makes the SFTP frontend
incompatible with the current stable release of Twisted. However it is closed
as a duplicate of #1525 because the improvement in #1525 would remove the
incompatibility. We briefly considered kludging it by requiring people to
install an older Twisted, but instead decided to fix it the good
way. David-Sarah had already implemented the fix to #1525 but there was
something wrong with the unit test.

NEWFLASH! This just in! As we were going to press, David-Sarah posted on
#1525 a comment that they are currently working on it.

We discussed #1767. David-Sarah had an idea for how to make the
implementation simple, by incrementing the server-wide counter once for each
service. Brian will work on it. It and #1732 are the two blockers that
require new code to be written before we release Tahoe-LAFS v1.10.

We closed #1484 as fixed. Yay!

#1746 is finished, reviewed, and ought to go into Tahoe-LAFS v1.10, but I
can't figure out how to merge it into master using git. I don't understand
how to use git very well, since I've only been using it on a daily basis for
a little more than two years now...

#1812 was already applied, but there was something wrong with it according to
Brian. Do we need to fix that as a blocker for 1.10 release?

We agreed to drop support for Python 2.5 and therefore for CentOS 5.

In the last few minutes of the hangout, we agreed that next week's hangout
will be another Nuts And Bolts like this one, hopefully bringing Tahoe-LAFS
v1.10 close to completion. Brian will work on #1767. Someone (maybe Zooko??)
will work on #1732.

Then we briefly mentioned the possibility of switching from a "semantic
versioning" style version number, to a YEAR.RELEASE_COUNT version number like
Twisted does. So instead of Tahoe-LAFS v1.10, this might be Tahoe-LAFS
v13.0. Then if there is another release in 2013, that would be Tahoe-LAFS
v13.1. We also considered leaving it alone for the "v1.10" release, and
switching styles for the next release.

Currently we use the "1." in "Tahoe-LAFS v1.10" to mean that it has
backward-compatibility with all of the stable Tahoe-LAFS releases since v1.0
(released March 25, 2008:
). I'm not sure that's accurate. While we have never deliberately violated
that goal, and we are not aware of any change that we've made which would
break that compatibility, on the other hand we don't have automated or manual
testing of backwards compatibility. Anyway, I suspect that there aren't any
users of Tahoe-LAFS versions older than 1.9.2:

tickets mentioned in this letter:

https://tahoe-lafs.org/trac/tahoe-lafs/ticket/867# use ipv6
https://tahoe-lafs.org/trac/tahoe-lafs/ticket/1484# CLI: overzealous
quoting of error messages
https://tahoe-lafs.org/trac/tahoe-lafs/ticket/1525# SFTP: handle
download failures correctly; remove use of IFinishableConsumer
https://tahoe-lafs.org/trac/tahoe-lafs/ticket/1679# Nondeterministic
NoSharesError for direct CHK download in 1.8.3 and 1.9.1
https://tahoe-lafs.org/trac/tahoe-lafs/ticket/1732# consider changes
to webapi "Move" API before release
https://tahoe-lafs.org/trac/tahoe-lafs/ticket/1746# write test for
anti-Ubuntu-crash-reporter exception-catching code
https://tahoe-lafs.org/trac/tahoe-lafs/ticket/1767# update
Announcement "timestamp": sequence number?
parse_abbreviated_size doesn't accept T for terabytes (and other
quibbles with the regex it uses)
https://tahoe-lafs.org/trac/tahoe-lafs/ticket/1926# Failed to load
application: cannot import name IFinishableConsumer


Zooko Wilcox-O'Hearn

Founder, CEO, and Customer Support Rep


More information about the tahoe-dev mailing list