Foundations Team Weekly Activity, 2009-07-17

Robbie Williamson robbie at
Fri Jul 17 15:16:43 BST 2009

== Present ==

 * RobbieWilliamson - chair
 * ColinWatson
 * EvanDandrea
 * JamesWestby
 * LarsWirzenius
 * MichaelVogt
 * MikeTerry
 * MuharemHrnjadovic
 * ScottJamesRemnant
 * SteveLangasek

== Apologies ==
 * MatthiasKlose - OEM sprint

== Agenda ==

 * Karmic
  * i486/i586 [Keybuk]
  * Status Updates on Alpha 3 Deliverables
 * Sponsorship queue
 * Objectives
 * Good news
  * Hardy and Jaunty are LSB 4.0 compliant :)
 * AOB

== Activity reports ==

=== Colin Watson ===
 * Removed nasty kernel version hardcoding from armel CD build scripts(the
replacement code is nearly as nasty, but at least shouldn't require frequent
 * Ran into udevd crash (#396957) and spent some time helping to track it down.
 * Reviewed/approved specs:
  * foundations-karmic-multiarch-support
 * Sponsorship:
  * libnice #392225.
  * python-pylibacl #391574.
  * scim-bridge #305452.
  * gparted #398099.
  * xscreensaver #398007.
 * Merged: dpkg, grub-installer, partman-auto
 * Archive admin half-day: cleared syncs (plus some minor tool fixes), did a
bunch of not-built-from-source removals, adjusted some kernel component
overrides, cleared priority-mismatches, cleared binary NEW
 * Feature work:
  * foundations-karmic-server-installer-improvements:
   * Finished and committed LVM configuration rearrangement.
   * Merged mdcfg into partman-md, partly because we've needed to do
     that for ages anyway and partly because it'll make the RAID
     configuration rearrangement much easier. Patch sent upstream for
   * Started work on RAID configuration rearrangement.
 * Adjusted finish-install for upstart 0.6.
 * Built i486/i586 live CDs out of live filesystems delivered by Adam.
 * Updated my parted swap naming patch; accepted upstream.
 * Reviewed relevant mail thread and added Muharem to the archive administration
team. Filed RT ticket for shell access.
 * Caught up on technical-board mail actions.
 * Fixed hdparm #396837 (crash when HDIO_DRIVE_CMD(identify) fails).
 * Fixed partman-auto #287571 (create_primary_partitions sometimes attempts to
create last partition larger than free space).
 * TB meeting.

=== Evan Dandrea ===
 * Investigated using wubi's tools/pywine to unit test the windows frontend and
backend for usb-creator.
 * Reviewed the oem-config and ubiquity merge a number of times,sharing input
and patches with mterry.
 * Dug at 397366, but have not found a suitable solution yet.
 * Merged UNR changes to Wubi for NCommander.
 * Undid the merging of bin/usb-creator-{gtk,kde}, cleaning up the code a bit
along the way.
 * Released a new Wubi and umenu for 8.04.3.
 * Sorted out some final travel details for Debcamp.
 * Debugged a bug in partman that turned out to be the fault of my desktop
running 9.04 rather than karmic.
 * Worked on the slideshow specification.  Mostly finished the ubiquity code.
Had a discussion with mpt about some aspects of the
design.  Tried to start a conversation with Gerry, but he's on holiday until
later this month.
 * Followed up with a community member (on ubuntu-installer@) who wanted an
option to not wipe a partition on install in ubiquity.

=== James Westby ===
==== Distributed Development ====
  * Importers down for 4 days due to LP API breaking. Back up now.
  * Cleaned up and improved the documentation somewhat.
  * Investigated some of the failures.

==== Daily Builds ====
  * No progress this week.

==== Kerneloops ====
  * Worked on the kerneloops changes.
  * Hit a snag that apport may not submit from the same machine that the problem
happened on, so need to be a bit more intrusive in to kerneloops than first thought.

==== Distro Tasks ====
  * Reviewed some packages on REVU.
  * "Half-day" archive admin duties on Monday.

==== Other ====
  * Chasing up confirmation for DebConf/sprint flights (again).

=== Lars Wirzenius ===

=== Michael Vogt ===
==== AppCenter ====
 * Code cleanup in gnome-app-install (AppCenter spec)
 * experiment with a apt-xapian-index plugin for the app-install-data-ubuntu
data, not happy yet
 * Performance analysis in g-a-i startup

==== software-properties ====
 * Bug triage/fixing in software-properties
 * Ensure daily apt cron job is executable when software-properties changes it
 * Port gnome-control-center to the new ubuntu-system-service
 * Software-properties: ensure only valid lines can be added with

==== synaptic ====
 * Integrate apt-xapian-index --update into synaptic
 * Debug problem with a-x-i and synaptic (constantly rebuilding), send patch to
 * Merge synaptic from debian, upload new version to ubuntu

==== unattended-upgrades ====
 * Add pm-utils hibernate hook to ensure that unattended-upgrades is finished
before going into final sleep
 * Merge unattended upgrades from the mvo branch, upload to karmic

==== update-manager/update-notifier ====
 * Show number of available updates in update-manager
 * Work on update-notifier to ensure that logrotated logs are considered for the
auto-launch code
 * Upgrade testing
 * Test update-manager landscape upgrade support and set spec to beta-available
 * Fix main-all auto-upgrade tester profile
 * Port update-notifier to use GAppInfo instead of g_spawn (#46258)
 * Fix crash in update-manager refresh_updates_count()
 * upload new  version

==== kernel-crashdump ====
 * update wiki because crash behaviour on sysreq-c changed in 2.6.31
 * work on the apport-retracer
 * Push apport changes for the kernel crashdump stuff to
 * add crash_signature support plus test
 * Discuss apport behaviour for local retracing (#392074)

==== apt ====
 * Work on more dynamic apt cdrom method(lp:~mvo/apt/libudev-dlopen)
 * Debug failure in main-all auto-upgrade test, turns out that for really huge
install lists there is a (hard to trigger) bug in libapt
 * Add check to apt.cron.daily to ensure that timestamps in the future get killed

==== misc =====
 * app-install-data-ubuntu update (review new desktop files etc)
 * Add new script "add-apt-repository" to
python-software-properties(ApturlPolicy spec)
 * Debug/workaround problem in gksu with the latest libgtk (with seb)
 * Make "include-links" the default in python-central and push to
 * improve spelling suggestions in command-not-found
 * Port ubuntu-system-serivce to polkit-1

==== sponsoring ====
 * computer-janitor, gnome-app-install
 * calctool (new ver), brasero (new ver), update-manager (bugfix)

=== Mike Terry ===
==== Ubiquity ====
 * Finally pushed the oem-config merge to trunk!  Thanks to Colin and Evan for
spotting and fixing some final issues.
 * Created a branch for translated timezones and filed a merge request. It's
very fancy, but there needs to be a bit of discussion about how to sort the
unicode lists.
 * Started looking at writing some unit tests

==== Rsyslog ====
 * MIR for rsyslog was approved
 * MIR for librelp was not, as there was a possible out-of-bounds error
in librelp that needs to be fixed.  I emailed the librelp folks about

=== Muharem Hrnjadovic ===
My Launchpad past caught up with me :P in the last week I was mostly sorting out
package set and archive permission issues that would stand in the way of the
archive reorganisation:
 * done: Fix hard-coded archive references in package set based archive
permissions (
 * in progress: Expose package set related archive permissions on the Launchpad
API; the 3 branches that implement this are currently being reviewed.
 * in progress: Package sets should be associated with a distro series; this has
been analysed for impact and a database patch was prepared but it's highly
unlikely these changes are going to land this week.

=== Scott James Remnant ===
 * i486/i586 performance comparisons; the result was that the big performance
gain we saw comes from gcc 4.4.  There is little changed between -march=i486 and
-march=i586 when we *also* force -mtune=generic as we do (generic is in fact
tuning for P4-era   processors).  In fact, the only thing enabled by -march=i586
-mtune=generic over -march=i486 -mtune=generic is the cmpxchg8b instruction,
which libc would use if it wasn't explicitly #ifdef'd out. It may be prudent to
do the switch anyway in case there are other optimisations later down the line.

==== Upstart ====
 * Upstart 0.6.0 was finally released!  This is a new stable version based off
the D-Busified 0.5 branch I'd been working on for the past couple of years in
between management duties.

 * After the release, was still having issues with the "expect fork" and "expect
daemon" code [which still uses ptrace()], debugged and applied a patch that
seems to fix it well enough until we move to the netlink code.

 * Created a "release" bzr branch of Upstart; this contains common history with
both upstart trunk and libnih trunk, merged together in the correct layout - and
then has the tarball unpacked over the top.

 * Created an Ubuntu packaging branch from that; I can "bzr pull" to update to
the latest release, and I can "bzr merge -c" to cherry-pick from trunk!  :-)

 * Some issues with building Upstart on ARM, the test suite triggers a compiler
bug and either hangs, runs out of memory or hits an Internal compiler error.  I
love my test suite.

==== udev ====
 * Updated to 145

 * Did some gardening of the bugs list, since I hadn't been focussing on it for
a while

==== util-linux ====
 * Got the branches all back in sync between Lamont and I
 * Added a commit for Colin's recent upload
 * Updated the Debian and Ubuntu trees to 2.15.1, uploaded to Ubuntu
 * Ported our hwclock udev rules to Debian and made the start rule of the init
script conditional on udev not running (agreed during Debian Boot sprint).

==== Kernel ====
 * The Upstart test suite strikes again!  This time it identified a regression
in inotify since it was entirely rewritten upstream to be based on fsnotify

 * Discussed the bug with upstream maintainer, it's being caused by merging
duplicate events for a directory based on only the mask - and not checking the
filename (ie. multiple CREATE, CHANGE or DELETE events for different files in
the same directory are merged into one for the first file)

 * Patches supplied to the kernel team since they just missed rc3

==== Misc ====
 * A fair amount of discussion over the gdm vs. getty bug, where both end up
sharing the same vt.  This is fundamentally caused by getty VTs being hardcoded
while X is free to chose the first it thinks is free. I'm still in favour of
moving getty to be not hardcoded, so our configuration would be "start N gettys"
and they'd get the lowest N numbered VTs after X

=== Steve Langasek ===
==== Release management ====
 * Ubuntu 8.04.3
  * finalizations of several pending SRUs
  * push fix for open-iscsi bug #236640
  * candidate images available with final kernel
  * image rerolls for wubi, umenu fixes
  * discussions with desktop team regarding bug #207072
  * identify and implement fix for bug #207072
 * phone call to discuss the escalation process for Support SRU bugs

==== Packages ====
 * clean up acpi-support dependencies

==== Specs ====
 * continued work with Debian on multiarch spec

==== Misc ====
 * Monday archive duties
 * Preparing DebConf talks
 * work with Dustin on pam_motd

Robbie Williamson                     		    robbie at
Ubuntu Foundations Team Manager             robbiew[]

"You can't be lucky all the time, but you can be smart everyday"
 -Mos Def

"Arrogance is thinking you are better than everyone else, while
Confidence is knowing no one else is better than you." -Me ;)

More information about the ubuntu-devel mailing list