For minutes of previous meetings, please see FoundationsTeam.
RobbieWilliamson - chair
- Release Manager Backup(s)
- 8.04.2 LTS Release (Jan 22)
- PDR Objectives
- Good News
Actions from this meeting
- Scott and Michael will backup Steve for the 8.04.2 LTS update
- Evan and Matthias will backup Steve for the Jaunty Alpha 3 and 4 releases
- Colin will prepare a summary of changes for the 8.04.2 LTS update
- Merged: mkvmlinuz, dpkg, xsane, base-files, hw-detect, parted
- Stable release updates:
- ubiquity #251751, #288479 (also rolling up other installer fixes,relevantly partman-base #149832, partman-target #224446, and user-setup #238224).
- Specification reviews:
- integration-best-practice (mostly OK, minor changes and comments)
- oem-tracking-id (substantial review comments)
- Minor seed fixes for encrypt-home-directory installer code. Advised kernel team on proper delivery of the relevant modules to the installer.
- ARM porting assistance:
- Fixed apex build failure.
- CD image script tweaks for mobile.
- UDS post-mortem call.
- Fixed tzsetup #28890 (jaunty bug-fixing, and needed for oem-config-server). Adapted oem-config to match.
- Discovered pygobject #309674 while investigating a ubiquity build failure.
- Regenerated hardy germinate output for Nick Barcet, and merged his supported seed split into intrepid and jaunty. Merged similar split for dapper.
- Finished first cut at oem-config debconf frontend.
- Wrote manual pages for all oem-config programs (#274048).
- Belatedly uploaded Totem cover image patch from Collabora (for BBC project).
- Wrote patch for partman-base #7928, sent upstream.
- Wrote patch for rescue #22311; needs to go upstream for translation support.
- Went through lots of old bugs, variously tagging them qa-jaunty-foundations or pet-bug, forwarding them, or closing them.
- Some end-of-year archive administration.
Fixed Task fields for packages in restricted and universe, and adjusted NewReleaseCycleProcess so that this doesn't get forgotten again.
- Tweaked seeds and debian-cd to offer OEM mode on server CDs.
- Minor bits of archive administration.
- Kernel patch to use 'git foo' rather than 'git-foo'.
- Platform/IS RT status call and notes write-up.
- Fixed mdcfg/partconf/partman-md #309555 (combination of ABI skew and fallout from modules being built into the kernel).
- Tested and followed up on lilo #221664, now fixed thanks to a Debian change.
- Followed up to ubiquity #290398 with patch to get more information.
Reviewed mobile -> netbook-remix seed patch from Steve Kowalik.
- Fixed usplash/gdm #304136 (console corruption at boot); hoovered up usplash #259761 and #313574 while I was there.
- Debate with kernel team and Scott about proper approach to breakage triggered by building lots of modules into the kernel (so 'modprobe foo' returns non-zero and the installer thinks features are missing). Worked around the rest of partman-crypto #314168 the hard way.
- Fixed partman-target #16871 (qa-jaunty-foundations). Hoovered up partman-target #221474 while I was there.
- Uploaded fix for partman-base #7928 from before Christmas.
- ltrace #313530.
- rootskel #309949 (sort of).
- autogen #313207 (archive administrator processing).
- user-setup #314482.
- Returned from holiday Jan 6
Some DistributedDevelopment discussions including a couple of sessions. Nothing too surprising as we already have an overall plan.
- The suggestion is to hang the Debian branches off the Ubuntu branches, as we hardly have any Debian history to import anyway, so it's not really worth re-writing history to include it.
- Need to write up a spec from the discussions and asses the impact on the overall spec that we have.
- Lot's of developer and community sessions, and loads of interesting discussions. There's a lot of great ideas being thrown around.
- Catching up on email and bugs, chasing a few issues.
- Booked travel to sprint and FOSDEM.
- Returned from holiday Jan 6
- Uploaded latest GNOME accessibility packages: at-spi, gnome-orca, mousetweaks, accerciser.
- Audio bug triaging.
- Merged Daniel Chen's pulseaudio changes into bzr.
- Resync local kernel tree to prepare for updating dm-raid4-5 module from usptream patch.
- Prepared a VM for kernel/userspace ddmraid testing in jaunty.
- Serious email catch up from over the break. Still need to reply to some.
- Track down some ARM related build failures, seen with the introduction of atomic helper functions in GCC.
- python2.4 and python2.5 updates, distutils prepared to accept a new --install-layout option to do nothing.
- OpenJDK updates for jaunty (openjdk6, cacao-oj6) and intrepid-proposed (openjdk6).
ARM buildd summary
- main looks pretty complete; the execptions are:
- acpica-unix: test failure
- compiz: still waiting for kdebase-desktop. Maybe build without KDE on armel until KDE is fixed to build?
fontforge: libtool problem, ArneGoetje working on a new upstream version.
- kdebase-workspace: ftbfs, required as a compiz b-d.
- kde*: build failures
- libapache2-mod-perl2: 2 failed tests
- libipc-sharelite-perl: 2 failed tests
- openoffice.org: missing b-d's in main
- rpm: combination of libtool and autoconf/automake doesn't work. no new qualified upstream version (not ARM specific). dependency of lsb.
- vbetool: maybe not supposed to build
- current status:
- Merges of packages updated in Debian.
- Forwarding of upstream bug reports.
- Python bug fixes.
- Email catchup.
Played with python-newt, wrote prototype for text (newt) based update-manager ui (http://people.ubuntu.com/~mvo/tmp/u-m-text2.png)
Intrepid->jaunty upgrade testing to verify bug that dholbach reported (vim) #296324
- experiment with the update-manager aufs branch (lp:~mvo/update-manager/aufs) some interesting results
- Add tests/ dir to synaptic and write some testcodde for it
- apt merges with debian
- python-apt work with potential new contributor
- SPECs work
- mail catchup after the vacation
- bug triage
- Work on the "codec install" spec implementation
- Upload new compiz/compiz-fusion-plugins-main
- Upload a bunch of packages for the libgnome-desktop transition
- read up about EC2 setup (tutorial)
- Fix FTBFS in gnome-main-menu
- Fix ftbfs in gnome-launch-box
- I18n merges in unattended-upgrades
- Apt-ddtp-tools work with sianis
- Write bzr cherrypick plugin that helps me keep branches in sync
discuss https://launchpad.net/aptdaemon with glatzor
- Catchup with compiz development and look at the new compiz++ branch
- Catchup call with robbie
- More work on gnome-codec-install and push to lp:~mvo/+junk/gnome-codec-install
- Review/test/sponsor gnome-settings-daemon
- Review/test/sponsor sysklogd sponsor request
- Review/merge gnome-desktop/gnome-control-center bzr branch
- usb-creator desktop file change (#286924)
- Fix icon size issue in update-notifier (#268085)
- Review/sponsor lifferea
- Reviewed/sponsored smartmontools
- Gnome-control-center update sponsoring and bzr mentoring
Scott James Remnant
- After much struggling, including hunting out and eventually writing a couple of patches to git(!!) and bzr fast-import (thanks Colin!) I got the udev git tree imported into bzr
- Created an ubuntu branch from the 124 release point on that and added the debian/ directory from each of the uploads to intrepid.
- Updated to 135 with a "bzr pull", and then spent the next little while completely redoing the packaging since 135 now uses autoconf!
- Working with Kay Sievers (udev upstream) to audit our udev rules, and try and merge everything into the upstream standard rules. This is still ongoing, but it really looks like we'll be able to ship our udev packages with *no* Ubuntu changes, not even to rules.d!
- Repackaged watershed into a separate source package, complete with udeb - this was the only remaining "extra"
- Did some instrumentation and script writing, making me able to do some performance analysis of udev 124 on the Mini 9:
- Compared to udev 135:
- Wrestled with some patches to make udevd multi-threaded, got them to apply in the end but couldn't quite get them to work! Mailed the author to see where I'd gone wrong.
- Imported the module-init-tools git tree into bzr, will be moving this packaging over to bzr as well:
- Opened discussion upstream about merging patches and modprobe.d files with the goal of having no differences here either (the kernel should ship its own modprobe.d files! so there :p)
- Asked the kernel folks to reduce CONFIG_LEGACY_PTY_COUNT to 0, have built a kernel with the same change for performance testing (numbers due soon)
- Posted my "notify init on reparent" patch to LKML:
http://lkml.org/lkml/2008/12/27/79 Some comments and resistance, but otherwise fairly quiet. Will address the race issue and resubmit as a patch.
Replacement laptop keyboard arrived, and I spent an exciting half an hour chasing a lost screw through the case while trying to fit it
- Performance Review calls with my old team members
- Wiki Gardening, did my required one hour.
- Finish merge of our rules with the upstream ones
- Check other standard rules in devmapper, lvm2, mdadm, ifupdown, etc. for potential necessary changes; esp. dep on watershed
- Continue performance testing.
- Update module-init-tools to latest upstream (with the binary index files).
- Packaging into bzr.
- Compare performance with existing modprobe.
- Cache answers to reduce subsequent lookup time and compare performance.
- Merge initramfs-tools with Debian.
- Packaging into bzr?
- Eliminate various modprobe calls based on what has a modalias already or what is non-essential and likely to be built into the kernel.
- Continue audit of system bus policy and usage.
- fix race condition in patch, address any other concerns, and resubmit
- Write specs.
- Merged Colin's bzr branch of uploads to jaunty, I got to nag him for not committing to bzr for once \o/
- Figured out again why we have a rule to set ID_TYPE and chatted with Colin about what the installer should do instead of relying on it.
- Worked through the remaining udev rules with Upstream, everything is now merged and our next udev packages will have no Ubuntu-local rules or patches!
- Went through udev bugs, fixed a few more minor ones
- Fixed lvm package to depend on watershed
- Removed use of watershed in mdadm by switching to incremental assembly, which I've been too much of a coward to do until now - so far, no reports of problems
- Updated dh_installudev to account for the move of rules into /lib, and for the new default rule number (40); handles migration automatically
- Rebuild upload of wacom-tools and pilot-link for new debhelper
- Put together list of other packages that need changing
Discussed how to make modprobe succeed silently when the requested module is built-in to the kernel; needs kernel help since only it knows its own config -> module name mapping
- If we can get a list of built-in modules from the kernel we can generate empty aliases for them, alternatively the kernel should expose what modules (not drivers!) are built-in in /sys
- Much performance analysis of various changes:
udev 124 -> udev 136
- udev 136 with/without Alan Jenkins' multi-threading patches
CONFIG_LEGACY_PTY_COUNT 256 -> 0
linux 2.6.27-9 -> linux 2.6.28-3
linux 2.6.28-3 -> linux 2.6.28-4 (built-ins!)
See https://lists.launchpad.net/ubuntu-boot/ for the results.
- Re-installed Jaunty on laptop for various testing purposes
- Wrote talk abstract and speaker bio for FOSDEM
- Release management
- jaunty alpha-2
- SRU processing for 8.04.2
- prod a number of the bugs that have been waiting in verification-needed for some time already
- 8.04.2 critical-path SRUs for acpi-support,openoffice.org-dictionaries
- coordinate with Colin, Robbie on a contingency plan for the next month in the event I'm selected for jury duty
- flesh out the hotkey troubleshooting guide based on the UDS session