0708

Differences between revisions 1 and 4 (spanning 3 versions)
Revision 1 as of 2009-07-08 13:53:17
Size: 1151
Editor: 99-156-85-10
Comment:
Revision 4 as of 2009-07-10 15:44:54
Size: 9951
Editor: 99-156-85-10
Comment:
Deletions are marked like this. Additions are marked like this.
Line 48: Line 48:
 * Fixed udev-udeb to avoid udeb-inappropriate dependencies.
 * Feature work:
  * archive-reorganisation:
   * Continued work on initial package sets.
  * foundations-karmic-server-installer-improvements:
   * Committed cdebconf online help code upstream, and synced it into Ubuntu.
   * Poked around at LVM configuration rearrangement. Not quite ready to test yet.
  * foundations-karmic-grub2:
   * Tested various GRUB installation scenarios, paying attention to RAID and to installation in a partition rather than in the MBR.
 * Sponsorship:
  * xmlrpc-c #376133.
  * kdeartwork #394168.
  * ssl-cert #352157.
 * Prepared release meeting agenda.
 * Reviewed regression-potential bugs:
  * Pulled in patch for system-config-lvm #390421 (fails with "invoke-rc.d: unknown initscript, /etc/init.d/lvm2 not found").
  * Pulled in patch for libgcrypt11 #389053 (stack smashing on VIA chipsets).
 * Fixed libcap #375595 (libcap2 FTBFS for lpia).
 * Fixed open-iscsi #364616 (wrong iscsistart path in iscsi hook).
 * Archive administration:
  * Extended backport-source (archive admin script) to be able to cope with backporting from a pocket (e.g. intrepid-security) for #385605; then found that syncbugbot was broken *again* due to edge drift not
    yet accounted for in python-launchpad-bugs. Instead of messing about with python-launchpad-bugs I decided to write client-side versions of the sync/backport scripts that could use launchpadlib to do archive queries and bug closing. http://bazaar.launchpad.net/~ubuntu-archive/ubuntu-archive-tools/trunk/revision/39?compare_revid=36
  * Wrote a client-side replacement for process-removals, thereby decommissioning the last local code we had that talked to the Launchpad production database directly (as opposed to via launchpadlib, or using scripts maintained as part of the Launchpad source tree).
 * Fixed user-setup #395082 (ecryptfs failure causes system to be silently set up without sudo access).
 * Made some progress on apt-setup #172879 (unable to skip network access during installation), with fixes to debconf and debian-installer-utils. These may turn out to be sufficient; I'll find out when they propagate onto CD images.
 * Updated installer autologin code to work with new gdm.
 * Adjusted initramfs-tools to cope with new compcache/ramzswap kernel module.
 * Fixed unattended-upgrades #396134 (unattended-upgrade-shutdown crashed with SyntaxError in unknown()), since Michael was away for the evening by the time I saw it.
 * Merged: dpkg (twice), debian-installer-utils, debconf
 * Fixed a problem with mdadm's /lib/partman/check.d/07root_on_raid hook that was causing distracting log noise.
 * Fixed partconf #396560 (karmic alpha: cannot setup software raid).
 * Debugged and fixed kernel desync problems with 8.04.3 CDs.
 * Reviewed lp:~mterry/ubiquity/oem-config-merge.
Line 50: Line 83:
 * Discussion with Colin and Jerone on improvements to the timezone map in ubiquity.
 * Ubiquity bug triaging.
 * Created a test harness for usb-creator. This creates a fake DeviceKit-disks so we can ensure that we keep the scope to just usb-creator, allow us to test without modifying the system, and allow us build up the KDE frontend without having DeviceKit-disks on the system. I plan to do something similar for the Windows frontend and backend using Wubi's code to run Python inside WINE, so that I don't have to keep booting into Windows.
 * Merged in the Windows backend code to usb-creator.
 * Started replying to Andrew Sayers follow up email regarding getting involved in migration-assistant development and how he sees his project fitting in.
Line 52: Line 90:
==== Distributed Development ====
  * More than 25,000 imported branches now hosted on LP.
  * Fixed some importer bugs.
  * Restructured the importer to use a process pool for importing, making it more efficient.
  * Hitting LP issues with timeouts that are far too short. Currently being investigated.
==== Daily Builds ====
  * Wrote short page on LP requirements for discussion with that team.
==== Kerneloops ====
  * Discussed requirements with kernel team.
  * Uploaded merge from Debian, and (re-)forwarded our patches.
  * Started looking at what modifications were needed to kerneloops.
==== Distro Tasks ====
  * A few items sponsored.
Line 54: Line 105:
 * Continued work on prototyping new UI for Computer Janitor(lp:~systemcleaner-hackers/systemcleaner/new-ui)
 * Started benchmarking of zsync for downloading .debs for updates. No results for this week's report, look for them in next week's.
Line 56: Line 109:
 * OEM Rotation
Line 58: Line 112:
==== update-manager/update-notifier ====
 * Test/fix auto-upgrade-tester with the official s3 images
 * Fix update-manager problem being too stealthy (#369820)
 * Work on making sure that auto-launch and install unattended work better together (#388953)
 * Fix ftbfs in update-manager (gtkbuilder changes)
 * Add code to u-m to show new installs more clearly in the UI
 * Port update-notifier so that it does no longer uses libgnome
 * Port update-notifier away from hal to gnome-disk-utils
 * code cleanup in update-notifier (-Wall clean)

==== AppCenter ====
 * Finish aptdaemon packaging, upload to karmic and the aptdaemon-developers-ppa
 * Work on apt-xapian-index to make a incremental index update, send upstream

==== apt ====
 * Build special apt for oem with a backport of a possible hang fix
 * Trying to reproduce apt dying (bug #385144)
 * discuss python-apt threading problem with sp
 * Debug python-apt i18n problem
 * Fix apt http download error when the server hangs up suddenly
 * Upload fix for apt http download problem into PPA and oem PPA for testing

==== unattended-upgrades ====
 * Add --dry-run to command-not-found, better output logging, deal with pkgs on hold
 * work on unattended-upgrades bugs, fix a bunch, ensure that a shutdown is delayed until dpkg is finished, upload new version (this is going to be useful for the update-on-shutdown spec)
 * Add some usplash awarness to unattended-upgrades-shutdown
 * Merge fixes from cwjatson from unattended-upgrades (many thanks!)
 * Fix error in connect() in unattended-upgrades-shutdown (#396263)

==== misc ====
 * Analyize update failure from support team
 * Debug problem with gksu and forkpty(), revert, notify upstream (took a bit until I found the actual problem, send to debian as #535544)
 * Work on the linux-crashkernel spec, update required linux-meta changes
 * Debug problem with gdebi-gtk, turns out to be upstream vte (#388953), fix and send patch upstream
 * Amule app-install-data problem fixed (#368580) + SRU for jaunty
Line 60: Line 149:
==== Ubiquity ====
 * Lots of final testing and small fixes for my oem-config merge.
 * Got initial review comments from cjwatson
Line 62: Line 154:
 * vacation
Line 64: Line 157:
 * Fix a race condition between init and initctl where the goal, state and process information wouldn't match because the three property requests were separate non-atomic calls. (ie. implement auto-generated proxy support for the GetAll properties method)
 * Implemented support to turn-off auto start for initctl messages, since it makes little sense for Upstart
 * Worked out why signals weren't getting through, caused by D-Bus always setting the sender to the unique name not the well-known name (and the proxy having the well-known name). Finally figured out what David Zeuthen's being bitching about all this time ;) Made the proxy objects always keep track of the name owner (I had this code already, it just wasn't used except for inhibit-style code) Added test cases for using well-known names, since I was always using unique names in the test suite
 * Upstart now returns "Access Denied" when trying to set a read-only property rather than "No Such Method"
 * Finish up test cases of passing structures over the bus (used for process information for initctl)
 * Add wait arguments for Start, Stop, Restart, EmitEvent so that you can change a job or emit and event without waiting for it complete while still getting errors
 * Version the Upstart D-Bus interface
 * Rewrite initctl from scratch, I'm much happier with it now - the syntax and output is much more like the 0.3.x series again
 * New initctl manpage
 * Update security configuration to allow read-only access to the list of jobs, instances and their information over the bus
 * Start work on better utmp/wtmp handling
Line 66: Line 170:
==== Release management ====
 * Ubuntu 8.04.3 work
  * chasing bug #236640
  * follow up on ISO testing
  * late kernel SRU processing, to recover from a -proposed+-security collision
 * preparing Ubuntu 6.06 desktop EOL

==== Packages ====
 * fix rpm to not pull in recommends/depends that will never be used on a sane Ubuntu system

==== Misc ====
 * Monday archive duties
  * tweaks to new mass-sync script

For minutes of previous meetings, please see FoundationsTeam.

Present

Apologies

Agenda

Outstanding actions from last meeting

Actions from this meeting

Activity reports

Colin Watson

  • Fixed udev-udeb to avoid udeb-inappropriate dependencies.
  • Feature work:
    • archive-reorganisation:
      • Continued work on initial package sets.
    • foundations-karmic-server-installer-improvements:
      • Committed cdebconf online help code upstream, and synced it into Ubuntu.
      • Poked around at LVM configuration rearrangement. Not quite ready to test yet.
    • foundations-karmic-grub2:
      • Tested various GRUB installation scenarios, paying attention to RAID and to installation in a partition rather than in the MBR.
  • Sponsorship:
    • xmlrpc-c #376133.
    • kdeartwork #394168.
    • ssl-cert #352157.
  • Prepared release meeting agenda.
  • Reviewed regression-potential bugs:
    • Pulled in patch for system-config-lvm #390421 (fails with "invoke-rc.d: unknown initscript, /etc/init.d/lvm2 not found").
    • Pulled in patch for libgcrypt11 #389053 (stack smashing on VIA chipsets).
  • Fixed libcap #375595 (libcap2 FTBFS for lpia).
  • Fixed open-iscsi #364616 (wrong iscsistart path in iscsi hook).
  • Archive administration:
    • Extended backport-source (archive admin script) to be able to cope with backporting from a pocket (e.g. intrepid-security) for #385605; then found that syncbugbot was broken *again* due to edge drift not
    • Wrote a client-side replacement for process-removals, thereby decommissioning the last local code we had that talked to the Launchpad production database directly (as opposed to via launchpadlib, or using scripts maintained as part of the Launchpad source tree).
  • Fixed user-setup #395082 (ecryptfs failure causes system to be silently set up without sudo access).
  • Made some progress on apt-setup #172879 (unable to skip network access during installation), with fixes to debconf and debian-installer-utils. These may turn out to be sufficient; I'll find out when they propagate onto CD images.
  • Updated installer autologin code to work with new gdm.
  • Adjusted initramfs-tools to cope with new compcache/ramzswap kernel module.
  • Fixed unattended-upgrades #396134 (unattended-upgrade-shutdown crashed with SyntaxError in unknown()), since Michael was away for the evening by the time I saw it.

  • Merged: dpkg (twice), debian-installer-utils, debconf
  • Fixed a problem with mdadm's /lib/partman/check.d/07root_on_raid hook that was causing distracting log noise.
  • Fixed partconf #396560 (karmic alpha: cannot setup software raid).
  • Debugged and fixed kernel desync problems with 8.04.3 CDs.
  • Reviewed lp:~mterry/ubiquity/oem-config-merge.

Evan Dandrea

  • Discussion with Colin and Jerone on improvements to the timezone map in ubiquity.
  • Ubiquity bug triaging.
  • Created a test harness for usb-creator. This creates a fake DeviceKit-disks so we can ensure that we keep the scope to just usb-creator, allow us to test without modifying the system, and allow us build up the KDE frontend without having DeviceKit-disks on the system. I plan to do something similar for the Windows frontend and backend using Wubi's code to run Python inside WINE, so that I don't have to keep booting into Windows.

  • Merged in the Windows backend code to usb-creator.
  • Started replying to Andrew Sayers follow up email regarding getting involved in migration-assistant development and how he sees his project fitting in.

James Westby

Distributed Development

  • More than 25,000 imported branches now hosted on LP.
  • Fixed some importer bugs.
  • Restructured the importer to use a process pool for importing, making it more efficient.
  • Hitting LP issues with timeouts that are far too short. Currently being investigated.

Daily Builds

  • Wrote short page on LP requirements for discussion with that team.

Kerneloops

  • Discussed requirements with kernel team.
  • Uploaded merge from Debian, and (re-)forwarded our patches.
  • Started looking at what modifications were needed to kerneloops.

Distro Tasks

  • A few items sponsored.

Lars Wirzenius

  • Continued work on prototyping new UI for Computer Janitor(lp:~systemcleaner-hackers/systemcleaner/new-ui)
  • Started benchmarking of zsync for downloading .debs for updates. No results for this week's report, look for them in next week's.

Matthias Klose

  • OEM Rotation

Michael Vogt

update-manager/update-notifier

  • Test/fix auto-upgrade-tester with the official s3 images
  • Fix update-manager problem being too stealthy (#369820)
  • Work on making sure that auto-launch and install unattended work better together (#388953)
  • Fix ftbfs in update-manager (gtkbuilder changes)
  • Add code to u-m to show new installs more clearly in the UI
  • Port update-notifier so that it does no longer uses libgnome
  • Port update-notifier away from hal to gnome-disk-utils
  • code cleanup in update-notifier (-Wall clean)

AppCenter

  • Finish aptdaemon packaging, upload to karmic and the aptdaemon-developers-ppa
  • Work on apt-xapian-index to make a incremental index update, send upstream

apt

  • Build special apt for oem with a backport of a possible hang fix
  • Trying to reproduce apt dying (bug #385144)
  • discuss python-apt threading problem with sp
  • Debug python-apt i18n problem
  • Fix apt http download error when the server hangs up suddenly
  • Upload fix for apt http download problem into PPA and oem PPA for testing

unattended-upgrades

  • Add --dry-run to command-not-found, better output logging, deal with pkgs on hold
  • work on unattended-upgrades bugs, fix a bunch, ensure that a shutdown is delayed until dpkg is finished, upload new version (this is going to be useful for the update-on-shutdown spec)
  • Add some usplash awarness to unattended-upgrades-shutdown
  • Merge fixes from cwjatson from unattended-upgrades (many thanks!)
  • Fix error in connect() in unattended-upgrades-shutdown (#396263)

misc

  • Analyize update failure from support team
  • Debug problem with gksu and forkpty(), revert, notify upstream (took a bit until I found the actual problem, send to debian as #535544)
  • Work on the linux-crashkernel spec, update required linux-meta changes
  • Debug problem with gdebi-gtk, turns out to be upstream vte (#388953), fix and send patch upstream
  • Amule app-install-data problem fixed (#368580) + SRU for jaunty

Mike Terry

Ubiquity

  • Lots of final testing and small fixes for my oem-config merge.
  • Got initial review comments from cjwatson

Muharem Hrnjadovic

  • vacation

Scott James Remnant

  • Fix a race condition between init and initctl where the goal, state and process information wouldn't match because the three property requests were separate non-atomic calls. (ie. implement auto-generated proxy support for the GetAll properties method)

  • Implemented support to turn-off auto start for initctl messages, since it makes little sense for Upstart
  • Worked out why signals weren't getting through, caused by D-Bus always setting the sender to the unique name not the well-known name (and the proxy having the well-known name). Finally figured out what David Zeuthen's being bitching about all this time Wink ;) Made the proxy objects always keep track of the name owner (I had this code already, it just wasn't used except for inhibit-style code) Added test cases for using well-known names, since I was always using unique names in the test suite

  • Upstart now returns "Access Denied" when trying to set a read-only property rather than "No Such Method"
  • Finish up test cases of passing structures over the bus (used for process information for initctl)
  • Add wait arguments for Start, Stop, Restart, EmitEvent so that you can change a job or emit and event without waiting for it complete while still getting errors

  • Version the Upstart D-Bus interface
  • Rewrite initctl from scratch, I'm much happier with it now - the syntax and output is much more like the 0.3.x series again
  • New initctl manpage
  • Update security configuration to allow read-only access to the list of jobs, instances and their information over the bus
  • Start work on better utmp/wtmp handling

Steve Langasek

Release management

  • Ubuntu 8.04.3 work
    • chasing bug #236640
    • follow up on ISO testing
    • late kernel SRU processing, to recover from a -proposed+-security collision
  • preparing Ubuntu 6.06 desktop EOL

Packages

  • fix rpm to not pull in recommends/depends that will never be used on a sane Ubuntu system

Misc

  • Monday archive duties
    • tweaks to new mass-sync script

FoundationsTeam/Meetings/2009/0708 (last edited 2009-07-10 15:44:54 by 99-156-85-10)