Foundations Team Weekly Summary, 2009-03-18

Robbie Williamson robbie at
Mon Mar 23 17:29:28 GMT 2009

== Present ==

 * RobbieWilliamson - chair
 * ColinWatson
 * EvanDandrea
 * LarsWirzenius
 * LukeYelavich
 * MatthiasKlose
 * MichaelVogt
 * ScottJamesRemnant
 * SteveLangasek

== Apologies ==
 * JamesWestby - On Holiday

== Actions from previous meeting ==
 * None

== Agenda ==
 * Jaunty
  * Beta Freeze Tomorrow
Status Update]]
  * [[|LP Targeted bugs]]
  * [[|Bug Fixes
Team Bug Fixes]]
 * [[|Sponsorship queue]]
 * Karmic
  * [[|UDS Session Planning]]
 * Good News
 * AOB

== Actions from this meeting ==
 * None

== Activity reports ==
=== Colin Watson ===
  * Fought with the mess that is ubiquity #282756. I've fixed the cciss bug that
was part of this; the only remaining legible part is a Kubuntu crash which I've
passed on to the contributor of the affected code.
 * Fixed partman-lvm #341046 (with any luck the last of the partitioning races,
and indications are certainly a lot better now, although I'm not counting my
chickens just yet).
 * Worked on ubiquity oem-priority #290400/#335596. There's now an extra knob
one can tweak to keep extra non-language-pack packages installed, which is
tweaked by default on the DVD to keep a couple of major application-level
packages installed, and the worst case for blacklist generation is down from
14:37 to 0:10. (The previous best case sees the least improvement from this and
is now the worst case: 4:00 -> 3:30.) Dell report being happy with this now, so
these are off the oem-priority list.
 * Archive reorg catchup call with Jono and Daniel.
 * Put together a reasonably minimal test case for qt4-x11 g++ ICE on powerpc
and filed #342335 with it. Fought with qmake for several hours to get the one
extra compiler option needed into the right place.
 * Fixed partman-target #320872 (check for duplicate filesystem labels).
 * Fixed tasksel #150252 (removes all ubuntu-desktop dependencies without
confirmation dialog).
 * 9.10 requirements call (boot time).
 * Some initial progress on user-setup #23475 (Cannot create a user "håkan").
Filed a Debian bug about the adduser piece that needs to be fixed first.
 * Fixed debootstrap #188690 (does not create a PO template on build; non-urgent).
 * Fixed debootstrap #320188 (PATH not exported by default when used as init=
 * Merged: os-prober
 * Fixed tasksel #148078/#326501 (various problems with Ubuntu Studio task
 * Fixed base-passwd #296841 (root password set to empty by default; not a cause
for panic since it's always overwritten in normal installations but not really a
good idea these days) and shadow/user-setup #340841
(passwd/root-password-crypted not being honored).

=== Evan Dandrea ===
 * More testing of Wubi. Filed more bugs, investigated them with Ago.
 * Fixed a bug in ubiquity where the format warning was not immediately showing
up, cleaned up the surrounding code in the process.
 * Fixed a bug in ubiquity wherein the partitioning bar on the advanced
partitioning page was showing giant percentages.
 * Released a few new versions of ubiquity, usb-creator, and wubi.
 * Generated some screenshots of Jaunty for the marketing department.
 * Attended a call on oem-config requirements from OEM Services.
 * Fixed the bug in the installer (335376) where it would just not show disks
for partitioning that had mounted partitions.  This was a necessary change as we
cannot re-read the partition tables of disks that have mounted partitions, so I
worked around it by wrapping it in a debconf question.  I realize that this
isn't the most pleasing UI, so anyone running the installer in an environment
with mounted partitions should preseed partman/unmount_active accordingly.  Do
note that you can do this from the kernel command line.
 * Fix for 340929, start networking with ubiquity when in only-ubiquity mode.
 * Followed up on a number of bugs and committed a few fixes to ubiquity.
 * Fixed Windows Vista and Windows 7 support in Wubi.
 * Blogged about recent Wubi changes to encourage testing.

=== James Westby ===
 On Holiday

=== Lars Wirzenius ===
==== Done this week ====
 * Reviewed update-manager bugs, starting from the old ones.
 * Subscribed to update-manager bug mail.
 * Helped Michael Vogt test aufs upgrades.

==== Remaining from past weeks ====
 * Fix bugs: 288731, 277924.
 * Find and fix other bugs.

==== Plans for the coming week ====
 * Fix bugs: 288731, 277924.
 * Continue review of update-manager bugs.
 * update MOTU application wiki page to include all my UDW presentations

=== Luke Yelavich ===
=== Audio ===
 * Audio bug triaging
 * Found some commits from upstream for more hda audio hardware enablement.
 * Uploaded some more PulseAudio revisions, thanks to Dan Chen for his work

=== Accessibility ===
 * Uploaded GNOME 2.26 releases of at-spi, atk, gnome-orca, accerciser,
mousetweaks, and gok. There is a documentation bug with atk which causes it to
FTBFs, which I've reported upstream. Hopefully this will be fixed in the coming
days, so I can upload a fix.
 * Tracked down and fixed a couple of accessibility related bugs, particularly
the inaccessible administrative applications bug.
 * Worked out that a race condition is causing logout dialogs to be
inaccessible, and notified upstream after coming up with a test to show that
this is the case.
 * Started writing a spec for the replacement of gnome-speech witgh
speech-dispatcher in future versions of Ubuntu, which will also involve work to
upstream code to more tightly integrate speech-dispatcher for use with GNOME and
 * Started to work on a BrlTTY related bug reported via mailing list from the

=== dmraid ===
 * Pulled in a couple of patches from Debian to fix some dmraid bugs relating to
intel software RAID controllers, and Promise controllers.

=== Misc ===
 * CD testing for UbuntuStudio, and accessibility testing for Ubuntu

=== Matthias Klose ===

=== Michael Vogt ===
==== update-manager ====
 * Fix crash in update-manager in checkFreeSpace,
 * fix deprecation warning in time.time()
 * upload new version to jaunty
 * Upgrade testing (manual and automatic) for alpha6
 * Write up post about aufs upgrade feature and ask for feedback
 * Fix crash in update-manager-text
 * Debug/fix bug in update-manger aufs upgrader (is_submount() code)
 * Fix translation issue in update-manager, upload new version
 * Diagnose aufs failure (#342451) and write patch for dpkg
 * Fix bug in update-manager with missing extra error catching on cache.update()
 * Debug/add extra paranoia to update-manager when removing obsoletes to fix
corner case (#335154)
 * do not overlay /var/cache/apt/archives in aufs mode

==== update-notifier ====
 * do not auto launch update-manager if the dpkg lock is already taken
 * Work on update-notifier logging/Fix update-notifier debug log output and
support debug logs for updates, hooks and inotify

==== compiz ====
 * Merge compiz apport hook update
 * add auto-adding gnomecompat if it is missing
 * Debug/fix UXA slowness with compiz via "sync_to_vblank" option
 * Debug compiz fade problem with macslow
 * Upload new compiz with fade/osd-notify patch

==== misc ====
 * bug triage
 * upload fix for update-initramfs problem (#338563)
 * debug python-ropemacs failure that turns out to trigger a python-central bug
(#341014), upload fix
 * upload new python-central that fixes that versions with no python-version
field can not be removed
 * Work on gnome-panel add-indicator applet migration and put a update into my
PPA (LP: #340777)
 * Upload python2.6 fix for gadfly (#314057)
 * Debug/fix crash in gdebi
 * Upload new gksu/libgksu
 * Command-not-found data update/review/upload
 * App-install-data review/update
 * Add new (debian) sections to synaptic
 * Debug/fix adobe-flashplugin meta package upgrade issue with the OEM team
 * Re-add codec install warning

==== sponsoring ====
 * Review/commit HIG fix for gnome-control-center (#327465)
 * Review/sponsor gnome-control-center upload
 * Review/sponsor vte, gnome-terminal, brasero

=== Scott James Remnant ===
 * Updates to bootchart packaging, repackaged the Java chart generator because I
prefer its output to the Python one
 * module-init-tools updated to 3.7-pre9
 * udev updated to 140
 * debugging the ehci_hcd warning about being loaded first; we now understand
why (see below [1]) and have decided to build the various USB host controller
drivers into the kernel and deal with it by link order
 * initramfs-tools fixed to include the kernel modules.order file
 * some comparison work on sreadahead vs. readahead-list with the different
possible configurations on both SSD and HDD
 * threw open the question of why X takes longer to start on Ubuntu than on Moblin
 * Created various bootcharts for call with Mark about boot performance
 * Installed Moblin Alpha 2 for boot comparison, notes supplied
 * Boot performance call with Mark
 * UDS pre-planning call
 * Discussion about /proc/bus/usb, uploaded fixes for fxload and
midisport-firmware; apcupsd was proven to be not broken; kvm and qemu are fixed;
and usbview has been requested for removal
 * Made a start on Upstart 1.0 work; the really boring bit of making the D-Bus
bindings fully functional to include properties, make asynchronous method calls
(with timeouts) and be able to catch signals - this will take a week or to, but
is worthwhile ground work

[1] ehci_hcd supports the USB 2.0 host, which communicates with USB 2.0 devices.
uhci_hcd and ohci_hcd support the two different possible USB 1.0/1.1 hosts,
which communicate with USB 1.0/1.1 devices.
The future xhci_hcd supports the USB 3.0 host, which communicates with USB 3.0
devices (though it may also support the 2.0 host and 2.0 devices). The host is
not the ports. If your computer supports both USB 1.0/1.1 and USB 2.0, you will
have two USB host controllers:
 1. 00:02.0 USB Controller [0c03]: nVidia Corporation CK804 USB Controller
[10de:005a] (rev a2) (prog-if 10)
 1. 00:02.1 USB Controller [0c03]: nVidia Corporation CK804 USB Controller
[10de:005b] (rev a3) (prog-if 20)
'''Note:''' prog-if 0x00 is uhci_hcd, prog-if 0x10 is ohci_hcd, and prog-if 0x20
is ehci_hcd <<BR>>
These hosts may share the same ports on the computer. This is why the load order
matters.  In order for USB 2.0 devices to work at full speed, the ehci_hcd
driver has to be loaded first to ensure it gets first refusal of them. Since the
hosts are separate PCI devices, no one host loads both modules, this is the kind
of thing that's very difficult to do in userspace. When we find the USB 1.0/1.1
host, we don't know that there's a 2.0 host yet to come.  The USB 2.0 host could
be on a PCI card or something that'll be initialised later. The rule basically
becomes "always load all three HCD drivers in a set order for any USB host" -
and at that point, we may as well just build them into the kernel. So that's
what we're doing.

=== Steve Langasek ===
==== Release management ====
 * alpha-6 released
 * FFe processing
 * continue chasing bug #328035 in the Intel driver

==== Specs ====
 * make grub2 available as an option in d-i expert mode

==== Packages ====
 * fix hplip-data, which was shipping .pyc files in the package
 * pull in new upstream version of freetype, to fix some crasher bugs and
rendering bugs

==== Misc ====
 * Monday archive duties

More information about the ubuntu-devel mailing list