Planning

Differences between revisions 35 and 66 (spanning 31 versions)
Revision 35 as of 2010-04-28 08:30:02
Size: 6396
Editor: eth0
Comment: fixed typo
Revision 66 as of 2010-05-03 14:31:07
Size: 14374
Editor: barry
Comment:
Deletions are marked like this. Additions are marked like this.
Line 1: Line 1:
<<TableOfContents>>

== Introduction ==

This page is specifically for the planning of the Maverick development cycle. As such, it initially included a number of links for proposed UDS sessions.

However, prior to UDS, we needed to organize the proposed Foundations track sessions, so this work was moved to a separate UDS planning wiki page: [[/UDSBrussels]]. The format of that page is simply a long list of sessions that can be ordered, adjusted, etc. This will help in that we will only maintain one wiki page with a complete list of blueprint links, not two (or more).

''Please note'': if you wish to add links to various blueprints for inclusion in the session planning efforts, '''do not''' put them on the page that you are currently viewing. Please add them instead to [[/UDSBrussels]].
Line 8: Line 18:
   * Have a UDS session for each track that is impacted by Upstart (e.g., Desktop, Server, Kernel, NewCore)
Line 10: Line 19:
   * [[UbuntuSpec:desktop-maverick-upstart|Desktop and Upstart]]
   * [[UbuntuSpec:server-maverick-upstart|Server and Upstart]]
   * [[UbuntuSpec:kernel-maverick-|Kernel and Upstart]]
   * [[UbuntuSpec:arm-m-upstart|Ubuntu on ARM and Upstart]]
Line 12: Line 25:
 * make screenshots.ubuntu.com show ubuntu branding and add support for PPAs and archive.canonical.com (MichaelVogt)  * make screenshots.ubuntu.com show ubuntu branding and add support for PPAs and archive.canonical.com (MichaelVogt) -- should this be a bug ticket? a work item in some blueprint?
Line 17: Line 30:
 * [[UbuntuSpec:foundations-m-uefi-support|UEFI Firmware Support]]
 * [[UbuntuSpec:foundations-m-openjdk-updates|OpenJDK6 updates for older Ubuntu releases]] (MatthiasKlose)
 * [[UbuntuSpec:foundations-m-cd-build-speed|CD build speed]] - CD build speed has been creeping back up of late; attack it again
 * [[UbuntuSpec:foundations-m-upgrade-and-install-testing|Discuss testing]]
Line 23: Line 41:
 * As part of their UNE work, Desktop may end up promoting Chomium to main  * [[UbuntuSpec:desktop-maverick-desktop-application-selection|Desktop Application Selection]] - As part of their UNE work, Desktop may end up promoting Chromium to main; may need help from Foundations to do this
Line 26: Line 44:
   1. goal: figure out how to deliver opportunistic apps via the software center
   1. finish ratings
 * Big push for touch, not sure what the impact is going to be for Foundations yet, but we need to be aware of it
 * It would really help Desktop if Upstart+Plymouth work stabilized so they could work with it
   1. [[UbuntuSpec:desktop-maverick-opportunistic-apps-stable-release|Opportunistic Apps in Software Center]] - figure out how to deliver opportunistic apps via the software center
   1. [[UbuntuSpec:desktop-maverick-sofware-center-ratings|Software Center Ratings]] - finish ratings work
 * [[UbuntuSpec:desktop-maverick-touch|Touch Support]] - Big push for touch, not sure what the impact is going to be for Foundations yet, but we need to be aware of it
 * [[UbuntuSpec:desktop-maverick-easy-wayland-testing|Wayland Testing]] - functional testing and related packages in Universe
Line 33: Line 51:
 * Wayland (no idea what state this is in upstream, and the X side of things should be desktop, but at the very least we'll need to look at how this overlaps/interacts with Plymouth);
Line 41: Line 58:
The server team has listed their planning by priority; their bullet items below reflect this.

 * Priority 1: Cloud cluster file system (not sure if we'll be needed for this or not)
 * Priority 2: Java/JDK improvements (doko does a lot of that; Server team is getting a Java person, so this may impact us less)
 * Priority 2: Support of alternate install modes
 * Priority 3: Upstart conversion
 * [[UbuntuSpec:server-maverick-cloud-gluster|Gluster FS for the Could]] - Cloud cluster file system (not sure if we'll be needed for this or not)
 * Java/JDK improvements (doko does a lot of that; Server team is getting a Java person, so this may impact us less)
   * [[UbuntuSpec:server-maverick-jboss|Interate JBoss]]
   * [[UbuntuSpec:server-maverick-java-frameworks|Java Frameworks]]
 * Support of alternate install modes
 * [[UbuntuSpec:server-maverick-upstart-conversion|Upstart conversion]]
Line 53: Line 70:
   * [[UbuntuSpec:dx-m-multi-touch-overview|Multi-touch overview]]
   * [[UbuntuSpec:dx-m-multi-touch-architecture|Multi-touch architecture]]
   * [[UbuntuSpec:dx-m-multi-touch-and-kernel|Multi-touch and Kernel]]
   * [[UbuntuSpec:dx-m-multi-touch-userland-support|Multi-touch Support in Userland]]
   * [[UbuntuSpec:dx-m-multi-touch-development|Mutli-touch Toolkit Development]]
   * [[UbuntuSpec:dx-m-multi-touch-app-conversion-and-dev|Multi-touch Application Conversion and Development]]
Line 56: Line 79:
 * TBD ==== General Priorities ====

 * Software Center enhancements (High) - Already planned
 * Multi-touch tablet support (Med) - Foundations work here may not be very strong; DX and Desktop, however, will be deeply involved

==== Base Platform ====

 * Hardware support for the new Intel (and likely AMD) CPU/chipset/GPU support (High) - 90% Kernel ownership; 10% Foundations
 * Remove use of intramfs in desktop and netbook editions (Med) - Foundations will support OEM removing this; however, Foundations actually needs initramfs to support more complicated use cases
 * Use swap files by default (instead of swap partitions) (Med) - Foundations would like to see this; there is an [[UbuntuSpec:foundations-karmic-swapfile|existing spec]]; currently blocked on a [[Bug:386554|kernel problem]] (out of memory)
   * More flexible than swap partitions -
   * Need to make this work with hibernate/hybrid suspend
 * Energy Star compliance power settings by default (High) -
 * Add btrfs support - option at install time (High) - Already planned
   * More efficient for storing lots of small files. Helps w/ OS size and may help boot time
   * Supports compression. Good for OS size and may help boot time too.

==== Usability and Performance ====

 * Continue boot performance work, 10s goal for Dell Mini w/SSD, Lucid is still above that (High) - This goal was for unity, however there is no harm in continuing work; most work on this will likely be done with Scott finishing upstart
 * Hybrid suspend/hibernate by default. Allows the system to go to sleep, but also stores a hibernate image on disk. If the battery dies, the system wakes from the hibernate image (Med) - this might be ideal... however, not sure how feasible it will be; would make suspend a lot slower; means we'd be forced to use dodgey kernel hibernate code

==== Internationalization ====

 * Collaborate with i18n upstream more closely, especially on release schedule, e.g., we missed ibus-1.3 for LTS 10.04 (High) - general platform issue; Foundations will host a UDS discussion for this
 * Per-language ISOs (e.g. a Chinese ISO, a Swahili ISO, etc) (Low) - Foundations can't support this as a whole -- it would explode our test cases; however, Foundations will work with OEM if they need help generating their own ISOs

==== Small Disk Configuration Support ====

 * Audit packages and see if "depends" can become "recommends", making it easier for OEM to reduce dependency lists. (High) - let's use the bug-filing process to accomplish this: OEM can file, and Foundations can address this on a case-by-case basis
   Example: xserver-xorg depends on xserver-xorg-video-all (actually, it depends on this OR a single driver) which pulls in *all* video drivers; if it were only recommended, you could specify only xserver-xorg-video-intel e.g. for your image
 * Remove all on-disk help/documentation and move it online. At the very least, split documentation into separate foopackage-doc.deb and make it a recommends. (High) - this is not appropriate for all use cases; we need to address this on a case-by-case basis; doing for all would guarantee that the index files would explode in size
 * Compressed r/w filesystem (High) - we will get this via btrfs
   * If btrfs is not feasible for some reason, we need a 'blessed' (by distro) r/w filesystem that supports compression (reiser, something FUSE-based, something else?)
   * We need r/w support to prevent the problem of updates taking too much disk space when using a r/o compressed filesystem like squashfs+unionfs
 * Updates for small disks - perform update as a sequence of partial atomic steps to minimize the use of disk (Med) - we do this now... so Foundations needs more detail to understand what need is not being fulfilled; OEM may mean downloading chunks and then applying it?

==== Other ====

 * BIOS validation tool (kernel team problem, not foundations? Colin King already working on this?) (High) - mostly Kernel with some Foundations
 * More stress testing before release (Med) - we will create UDS session for this: improving our testing procedures, using more automated tests, further brainstorming, etc.


== Support FOR Foundations ==

Collect here items which we would like to see be done by other teams

=== Desktop ===

 * Java Applet maintainance / development
 * Stable xulrunner-1.9 interfaces when having more than one xulrunner version on the desktop (both eclipse and the icedtea6-plugin don't seem to get it right).

=== Launchpad ===

 * Soyuz/IS: Rebuild tests take too long, or slow down PPA buildds.
 * Soyuz: Native syncing via web service, to free up man-hours of archive admin time per week.
 * Buildd configuration, and how to reproduce results from test runs on the buildds. Differences in the configuration of distro and PPA buildds. LP: 556597 as an example. Seen with some eglibc test cases as well.

=== QA ===

 * TCK testing
Line 65: Line 148:
   * Automated transition testing framework (build, run upstream unit test, 2to3, report on ftbfs and test failures, against any/every python version)
   * Migration to Python 3?
   * Eliminate symlinks for python-[central|support] using PEP 3147?
   * Python 2.7?
   * Freezing (e.g. cx_freeze) critical dist upgrade tools?
   * [[UbuntuSpec:foundations-m-python-continuous-integration|Automated transition testing framework (build, run upstream unit test, 2to3, report on ftbfs and test failures, against any/every python version)]] (BarryWarsaw)
   * [[UbuntuSpec:foundations-m-python-3|Migration to Python 3]] (BarryWarsaw)
   * [[UbuntuSpec:foundations-m-python-no-symlinks|Eliminate symlinks for python-central and -support using PEP 3147? Needs support for extensions. (And mostly done for python-central).]] (BarryWarsaw). This is a duplicate of UbuntuSpec:foundations-lucid-robust-python-packaging.
   * [[UbuntuSpec:foundations-m-python-2.7|Python 2.7]] (BarryWarsaw) That should be tested before UDS as a basis for a decision.
   * [[UbuntuSpec:foundations-m-python-cx-freeze|Freezing (e.g. cx_freeze) critical dist upgrade tools?]] (BarryWarsaw)
 * [[UbuntuSpec:foundations-m-drop-ia64sparc|Drop the IA64 and SPARC community ports]]
 * Compile for i686 as the default on i386 (instead of i486).
 * [[UbuntuSpec:foundations-m-toolchain|Toolchain decisions for Maverick]]. May be followed up with architecture specific sessions. (MatthiasKlose)

Introduction

This page is specifically for the planning of the Maverick development cycle. As such, it initially included a number of links for proposed UDS sessions.

However, prior to UDS, we needed to organize the proposed Foundations track sessions, so this work was moved to a separate UDS planning wiki page: /UDSBrussels. The format of that page is simply a long list of sessions that can be ordered, adjusted, etc. This will help in that we will only maintain one wiki page with a complete list of blueprint links, not two (or more).

Please note: if you wish to add links to various blueprints for inclusion in the session planning efforts, do not put them on the page that you are currently viewing. Please add them instead to /UDSBrussels.

Ideas

Foundations Support

As usual, Foundations will be providing Ubuntu support for the other teams. A high-level listing of projects where we likely will be involved is given in this section.

Desktop

Design

  • We probably ought to have a look-in on the multi-screen work, if it's dealing with remembering configurations for individual devices;
  • Installer, first boot, upgrades, Update Manager, and Software Centre;
  • System settings (maybe, depending on what this comes down to);
  • Possibly keyboard indicator but it depends what they're doing (if it's just visual then probably not).

Server

Kernel

OEM

General Priorities

  • Software Center enhancements (High) - Already planned
  • Multi-touch tablet support (Med) - Foundations work here may not be very strong; DX and Desktop, however, will be deeply involved

Base Platform

  • Hardware support for the new Intel (and likely AMD) CPU/chipset/GPU support (High) - 90% Kernel ownership; 10% Foundations
  • Remove use of intramfs in desktop and netbook editions (Med) - Foundations will support OEM removing this; however, Foundations actually needs initramfs to support more complicated use cases
  • Use swap files by default (instead of swap partitions) (Med) - Foundations would like to see this; there is an existing spec; currently blocked on a kernel problem (out of memory)

    • More flexible than swap partitions -
    • Need to make this work with hibernate/hybrid suspend
  • Energy Star compliance power settings by default (High) -
  • Add btrfs support - option at install time (High) - Already planned
    • More efficient for storing lots of small files. Helps w/ OS size and may help boot time
    • Supports compression. Good for OS size and may help boot time too.

Usability and Performance

  • Continue boot performance work, 10s goal for Dell Mini w/SSD, Lucid is still above that (High) - This goal was for unity, however there is no harm in continuing work; most work on this will likely be done with Scott finishing upstart
  • Hybrid suspend/hibernate by default. Allows the system to go to sleep, but also stores a hibernate image on disk. If the battery dies, the system wakes from the hibernate image (Med) - this might be ideal... however, not sure how feasible it will be; would make suspend a lot slower; means we'd be forced to use dodgey kernel hibernate code

Internationalization

  • Collaborate with i18n upstream more closely, especially on release schedule, e.g., we missed ibus-1.3 for LTS 10.04 (High) - general platform issue; Foundations will host a UDS discussion for this
  • Per-language ISOs (e.g. a Chinese ISO, a Swahili ISO, etc) (Low) - Foundations can't support this as a whole -- it would explode our test cases; however, Foundations will work with OEM if they need help generating their own ISOs

Small Disk Configuration Support

  • Audit packages and see if "depends" can become "recommends", making it easier for OEM to reduce dependency lists. (High) - let's use the bug-filing process to accomplish this: OEM can file, and Foundations can address this on a case-by-case basis
    • Example: xserver-xorg depends on xserver-xorg-video-all (actually, it depends on this OR a single driver) which pulls in *all* video drivers; if it were only recommended, you could specify only xserver-xorg-video-intel e.g. for your image
  • Remove all on-disk help/documentation and move it online. At the very least, split documentation into separate foopackage-doc.deb and make it a recommends. (High) - this is not appropriate for all use cases; we need to address this on a case-by-case basis; doing for all would guarantee that the index files would explode in size
  • Compressed r/w filesystem (High) - we will get this via btrfs
    • If btrfs is not feasible for some reason, we need a 'blessed' (by distro) r/w filesystem that supports compression (reiser, something FUSE-based, something else?)
    • We need r/w support to prevent the problem of updates taking too much disk space when using a r/o compressed filesystem like squashfs+unionfs
  • Updates for small disks - perform update as a sequence of partial atomic steps to minimize the use of disk (Med) - we do this now... so Foundations needs more detail to understand what need is not being fulfilled; OEM may mean downloading chunks and then applying it?

Other

  • BIOS validation tool (kernel team problem, not foundations? Colin King already working on this?) (High) - mostly Kernel with some Foundations
  • More stress testing before release (Med) - we will create UDS session for this: improving our testing procedures, using more automated tests, further brainstorming, etc.

Support FOR Foundations

Collect here items which we would like to see be done by other teams

Desktop

  • Java Applet maintainance / development
  • Stable xulrunner-1.9 interfaces when having more than one xulrunner version on the desktop (both eclipse and the icedtea6-plugin don't seem to get it right).

Launchpad

  • Soyuz/IS: Rebuild tests take too long, or slow down PPA buildds.
  • Soyuz: Native syncing via web service, to free up man-hours of archive admin time per week.
  • Buildd configuration, and how to reproduce results from test runs on the buildds. Differences in the configuration of distro and PPA buildds. LP: 556597 as an example. Seen with some eglibc test cases as well.

QA

  • TCK testing

Placeholders for Big Topics

Lucid specs not done

(Numbers in parentheses indicate to-do/postponed/done.)

Lucid specs partly done

Lucid specs mostly done

FoundationsTeam/Maverick/Planning (last edited 2010-05-04 14:04:46 by p5B09F55E)