For minutes of previous meetings, please see FoundationsTeam.
RobbieWilliamson - chair
LarsWirzenius - Injured back
Actions from previous meeting
Robbie to organize EC2 demo - dendrobates is going to organize this for the Platform team during the sprint
- Jaunty (Alpha 4 release - Feb 5)
Bad News - Lars injured his back and cannot join us in Berlin
- Good News
- 8.04.2 LTS Release...good work Steve!
Actions from this meeting
- Fixed Ubuntu 8.04.2 CD build problems (#253255 and an unfiled jigdo-generation bug that showed up in log files).
- Spec ready for review.
- Further work on adding manual package selection to the server installer. Ran into and fixed an amusing cdebconf plugin handling bug that seems likely to delay Debian's Lenny release in order to squeeze in a fix (whoops). Stalled on manual package selection for the moment in favour of getting other things done, though; more at the sprint.
- More progress on spec; getting there.
- Fixed debconf bugs filed in Debian as #476873 and #496631, which broke oem-config-server.
- Some more spec work.
- Brief update at Technical Board meeting.
- Discussion with Scott Kitterman about MOTU organisation.
- Debugging of netcfg #8980 (qa-jaunty-foundations); the remainder appears to be mostly a network-manager bug.
- Started intrepid SRU process for busybox #293586.
- Merged Debian's grub/grub-installer XFS patches, fixing #8058.
- Enabled stty in busybox for Oliver. Fixed busybox build failure discovered as a result of this.
- Fixed poor documentation of Kickstart preseed command (an Ubuntu extension) which didn't properly describe quoting rules.
- Fixed libvte9-udeb's dependencies to fix d-i builds (#321848).
- Fixed partman-crypto #321732 (bad crypttab when using key types other than passphrase).
- lintian #318212.
- user-setup #307443.
- bash-completion #316654.
- antlr #318362.
- tiger #318933.
- lm-sensors-3 #321632.
- app-install-data/thunderbird-quickfile #313564.
- No-change upload of evolution to get it built on armel.
- Reviewed "Automated deployments of Ubuntu" white paper for accuracy; sent notes to Nick Barcet.
Modified mksquashfs to reduce progress bar noise in logs (live filesystem log files were >7MB almost entirely due to this!). Sent patch upstream.
- Lots of bug triage and archive administration. Attempted to catch up on several months' backlog in my =ubuntu/mybugs mail folder; down from 27000-odd to 9000-odd ... but feeling much more up-to-date now.
Short week, swap day on Friday.
- Fixed umenu invalid CD issue.
Had a long back and forth with Rudd-O, the developer of Portablelinux (https://edge.launchpad.net/portablelinux). He wanted to merge usb-creator into portablelinux, but we had very different use cases and design considerations, so I respectfully declined.
Modified the keyboard layout page in ubiquity to adhere to the JauntyUbiquityUsability specification. Some discussion with mpt in Millbank and cjwatson online over the design and where to go from here.
- Discovered bug 318564. Found a workaround, but am still investigating the root cause of the bug and a proper solution.
- Started looking into porting the keyboard detection code from cdebconf-keystep to ubiquity.
- Finished a basic cut of the new timezone map. Need to profile and optimize next, ask Ken for new, more accurate artwork, and integrate it into ubiquity. Need to determine which timezones must be included in the map, and which can be relegated to the drop down menus.
Polished up SegmentedBarSlider for inclusion in ubiquity. Still needs a pretty resize handle and a slight reworking of the percentage code.
- Uploaded a new user-setup that accommodates ubiquity's need to have ecryptfs-utils in user-setup-ask. Given this change and kirkland's most recent ecryptfs-utils upload, the next ubiquity upload should provide a working encrypted home option.
- Distributed Development
- Work on the Jaunty spec. The main difficulty is working out how to stitch the Debian history in to the Ubuntu one with incomplete information and without incorrect merge information. Can be done easily for most cases, but some packages may end up with branches with no common ancestor, which is troublesome.
- Deal with package-import.ubuntu.com moving to bzr 1.11 and a couple of inadvertent changes that came with that.
- Started porting the importer to launchpadlib, should massively decrease resource consumption.
- Discussions on improving various things about building packages using builddeb. Some ideas are starting to coalesce, but there are still some unanswered questions.
- Wrote a blog post based on my Developer Week session on packaging with bzr. It shows how things will look once everything comes together.
- Call with Jono and Daniel, see ubuntu-devel for meeting notes.
- Caught up with jml, launchpad work is progressing nicely, first pieces in place on staging.
- A couple of uploads, and a couple of things sponsored, not a great deal of activity.
- Preparing from sprint.
- Attended OEM/Platform meeting.
NOTE: National holiday on Monday January 26th.
- Audio bug triaging.
Found out that a quirk for the MacBook Pro 4,1 machines was submitted upstream to the alsa devs, but that quirk doesn't include my machine. Start investigating how to fix this properly, since adding my machine's audio subsystem ID is not enough sound work without the need for a module parameter.
- Continued to get fixes either from Daniel T Chen or upstream for pulseaudio.
- Pulled audio hardware enablement fixes from alsa git upstream, and made sure they compiled. Will send to the kernel team once these fixes are confirmed to compile for all amd64 and i386 kernels.
- Investigated an issue some users are having with not being able to run alsamixer in a 64-bit jaunty isntall. The only report has been from a fresh install so far, however Daniel T Chen and myself are unable to produce the error, although I was unable to run alsamixer after fetching updates a few times. More investigation needed.
- Prepared to do more dmraid installation with ubiquity testing on real hardware, to find that the video card in the machine containing a dmraid compatible IDE RAID card is not working properly with jaunty xorg. Still need to file a bug, but have gathered the needed bits for the report.
- Continued to debug an issue with the dmraid-activate script, relating to the attempted activation of intel software RAID arrays. Will need to special case the way that dmraid works with isw raid set names and groups in the activation script.
- Heard about efforts upstream to *FINALLY* combine mdadm and dmraid. It seems mdadm is getting support for working with metadata formats, and upstream already supports the intel software raid metadata format, which is wonderful news. I will keep my eye on this one closely, as this will hopefully bring sanity to the dmraid mess we currently have.
- No sponsoring. :S (Need to do it earlier in the week to be sure of more sponsor requests being open, they seem to mostly be satisfied once my time zone comes around.)
Sent a fix to the kernel team to incldue the now broken out hid modules. I attempted to install jaunty on my MacBook Pro, and had no keyboard during the alternate installer, and had to connect a USB keyboard.
- Met with Michael Hudson to talk about test/rebuild UI requirements in soyuz/lp.
- GCC-4.3.3 final release, test rebuilds, set optimization defaults to the already agreed defaults (cpu=armv5, tune=cortex8).
- python-2.6.1 updates.
- Phone call with Dalibor Topic about OpenJDK development.
- add option to allow unsetting the mirror check
- work (a bit) on the aufs branch
- Better development release announcements
- added branch that can render the wiki markup for the release announcements
- Update-manager mirror usage/detection improvements (with tests etc)
- code cleanup (python modules)
- better debug output
- upload new version into jaunty
Sru for dapper->hardy update-manager dist-upgrade failure on sparc (#239458)
- bug triage
- Work on the jaunty foundations buglist (#24250, #107983, #111819, #138687, #107983)
- attack the gdebi buglist and fix a bunch of them and uploaded gdebi 0.4.0
- Debug/fix bug in vmbuilder that can lead to a empty domainname in /etc/hosts (and that confuses postfix)
- report bugs on package that do not install/remove cleanly (nvidia, libboost1.37) in jaunty
- better detection for install failures (important for base test image building)
- isolate testcase for gst.Caps.intersects() vs. gst.Caps.is_subset() problem/fix and sent to upstream
- upload fixed gnome-codec-install
- fix crash due to wrong indention
- Port python-apt from using iso_3166.tab to iso_3166.xml for getting county names
- fix for socket.setdefaulttimeout,
- workaround for missing pep 341 try..finaly in 2.4
- Sru for python-apt (the other half of the upgrade failure for sparc #239458)
- prepare/give "fun with python-apt" presentation
- travel preparing berlin
- port software-properties to use iso_3166.xml and upload new version
- investigate zsync again and compare it for Packages.gz against pdiffs
- checking out allmyapps.com
initial merging back of ddtp translations to debian (http://ddtp.debian.net/ddtss/index.cgi/es is pretty impressive)
- examine slowdown of dpkg for old installs and adding a cruft-remover plugin that helps cleanup the dpkg status/availalbe
- experiment with the dpkg patch that avoids reading the available file (debian #397121) and put it into my PPA
- spec review language-selector
- Review/sponsor packagekit,packagekit-gnome uploads
- Review/sponsor brasero
- Patched to allow running without user/group lookup, using it now in
- initramfs and installer
- Put default rules into initramfs
- Removed MAKEDEV from /dev because FHS now says we can
- Diverted udevadm out of the way during udev upgrade to trap people
- running trigger or settle into depending on udev itself
- Rules fixes for hplip
- Discussion upstream about where we'll put the current catalogue of
- "vendor id" and "device id" matching rules that exist for both udev and in hal-info (udev-extras was the decision)
- udev 137 released and uploaded
- Started detailed work to get udev even faster, it's now 1.4s - we can do better (the last events are the disk ones, but they don't take that long, so other overheads must be delaying them)
- Patched to allow running without user/group lookup, using it now in
Set up a GIT->bzr import
- Packaged from scratch based on that import
- Discussions with upstream
- Merged all patches upstream (will be in 3.6) so we can ship patch-less
- Thoughts/experiments of shipping modprobe.conf with the kernel in /lib/modules/$(uname -r) - since many options are to work around kernel bugs/quirks
- Performance work on modprobe binary index
- Performance testing
- Boot performance
- Phone call and e-mail discussion with Ken Edwards
- Ubuntu Developer Week session prep
- Installed the Babbage Board
- Rewrote nih_alloc() to allow multiple references to objects, discovered a way of doing allocations that automatically free when the pointer goes out of scope unless the object was referenced in the meantime
- Ported libnih and Upstart to the new allocator
- 0.5.1 released
- Wrote the boot performance spec finally, now in Review
- Worked on my FOSDEM/BOSSA talk
- Fall-out from discovering that ARM doesn't support pselect() or ppoll()
- Have a kernel.org account now, where I'm publishing git repos to make getting any patches to udev, module-init-tools and util-linux-ng upstream
- Continued discussion with Ted Tso, Kay Sievers and Karel Zak about how we solve our vol_id/blkid differences; hoping Karel will merge his util-linux-ng branch any day now
- Did some investigatory work on how we deal with changes to block devices and invalidating udev's /dev/disk/by-* symlinks
- Patch to util-linux-ng to emit change event on any block device mkfs/mkswap are run on
- But then thought it might be better to do it in the kernel; so been reading the block device layer code and working out how to get a change event in there if the block device is written to from userspace
- Release management
- 8.04.2 release
- coordinate testing of a second round of ISOs, fixing a debian-cd bug when including updates-only packages (LP #253255)
- follow up on hw certification; one regression reported on non-certification hardware, LP #319840
- clean up milestone targets for any deferred SRUs
- follow through on incomplete SRUs that were inadvertently included in the 8.04.2 ISO images
- triaging of jaunty nominations
- 8.04.2 release
- continued analysis of login failures with pam_smbpass (bug #292791)
- follow through on nut SRU (bug #222761), to get it resolved
- fix sound problem with flashplugin-nonfree on amd64 due to a missing pulse dep in ia32-libs (bug #314739)
- fix up xrandr regressions in gnome-settings-daemon
- merge samba to fix a couple of release-critical bugs (#286119, #264943)
- fix libxcb to use Breaks: instead of Conflicts:, for cleaner upgrades
- missed this week; will do double-duty next week
- jaunty upgrade testing, pick apart some NM regressions
- Monday archive duties