Ubuntu Wiki

Please check the status of this specification in Launchpad before editing it. If it is Approved, contact the Assignee or another knowledgeable person before making changes.

• Launchpad Entry: composite-by-default.

• Packages affected: compiz-core gnome-control-center ccs-settings libccs python-ccs compcomm-plugins (no actual name as of yet...) emerald emerald-theme-manager gnome-keybinding-settings

See also:

• ["Xorg7.3Integration"] - Integrate Xorg 7.3 components, adopt config autodetection

• ["BulletProofX"] - Failsafe mode when X doesn't start, with displayconfig-gtk to configure; includes a whitelisting capability that CompositeByDefault could share

Summary

• Deploy Compiz, with a set of plugins and settings backends as the default window manager in Gutsy Gibbon.

Rationale

• Compiz is now able to function as a drop in replacement for Metacity. Due to increasing support for required GL extensions in both free drivers and proprietary drivers, and increasing stability in the project and it's assosciated plugins the software is usable and available for a broad base of users. User's have shown a large amount of interest in Compiz/Beryl for making their computing experience more usable and enjoyable. The highly visible effects of Compiz can act to encourage users to try Ubuntu.

Use Cases

• John is a new user coming from windows and finds the idea of workspaces/viewports completely foreign. The wall plugin for Compiz presents these to John in a tactile and intuitive way helping his workflow.
• Steve is a power user and often works with dozens of windows. Plugins such as Scale (tile out all windows and let you pick one) helps him to stay organized and work faster.
• Edward is visually impaired. The neg and zoom plugins help him to zoom in close on small text, and improve contrast in some interfaces.
• Mark likes wobbly windows and other special effects and wants to have them on his Ubuntu desktop.

Scope

• Deploying Compiz as the default window manager in Ubuntu involves changes in several areas. Work with Compiz will have to be done for integration in to the existing GNOME desktop, behavior should be consistent with metacity when possible (true to a degree already), and users should be able to both include their metacity settings, and use existing GNOME tools to configure Compiz (already true to a large degree again). libwnck will have to be patched (patch available, probably going in upstream) to properly support viewports. A theme for the Emerald decorator will have to be prepared, blending in with the Metacity human theme, but additionally including some of the effects allowed by compositing.

• Sane default plugin settings will have to be chosen, along with a sane default set of plugins such as to provide an impressive, but not overwhelming experience.

Design

• Integration with the GNOME desktop will be largely handled through the Compiz Configuration System (CCS). CCS is a plugin, library, set of backends, and tools for Compiz providing the following:
  • Consistent manner to access Compiz settings stored in GConf, ini (flat file), or kconfig backends.
  • Exporting, importing, and editing of profiles (For defaults, etc...) from/to any backend.
  • Python bindings around the library to enable easy manipulation of settings from Python apps.
  • Reading and writing of Compiz settings while Compiz is not running.
  • Translation in the GConf and KConfig backends from a large number of metacity/kwin settings to the equivalent Compiz settings, including watching for changes on the keys.
  • Support for multiple X Screens.

• Some applications such as gnome-keybinding will be patched to use CCS to add settings for Compiz specific keybindings. A default set of plugins, animations, etc is chosen, with the rest being shipped by default or available for the user to enable if they so desire. Desktop effects will be moved in to GNOME apperance properties.

Implementation

1. Ensure all required components are present in main and up to date with upstream
   • compiz-core - Need to discuss with upstream about producing a release; otherwise we will need to use a git snapshot
   • different plugins packages (one source package multiple binary packages?)
   • CCS Plugin/Library.
   • CCS Python bindings.
   • CCS Backends.
   • CCS Settings Managers.
   • CCS-GNOME/CCS-KDE metapackages.
   • emerald
   • emerald-theme-manager
2. Establish "standard plugin list":
   • move
   • resize
   • place
   • snap
   • wall
   • expo
   • zoom
   • neg
   • video
   • animation
   • switcher
   • fade
   • ccp/gconf
   • dbus
   • decoration
   • extrawm
   • jpeg, png, svg (image loader plugins)
   • regex
   • text
   • scale
3. Establish "optional plugin list" (checkbox to turn on):
   • ring (replacing switcher with same keybindings)
   • trailfocus
   • blur
   • wobbly
4. Establish a whitelist of Drivers/PCI ID's where composite can be enabled by default
   • See ["BulletProofX"], which will be maintaining a similar whitelist that could be reused here
5. Integration with existing GNOME desktop
6. Patch libwnck to properly support viewports (patch exists, may be incorporated upstream).
7. Create theme for the Emerald decorator
   • Strive to make it blend in with the Metacity human theme
   • Incorporate the following composite effect animations:
     • glide 1 for create Transients use fade
     • glide 2 for close Transients use fade
     • minimize/maximize zoom
     • shade - rollup
     • Focus - fade
   • Identify sane default plugin settings to craft an impressive but not overwhelming experience

Test Plan

• Evaluate behavioral differences between Metacity and Compiz
  • For each difference identified, determine whether to keep Compiz behavior or adopt Metacity behavior
• Verify that metacity window manager settings are correctly used by Compiz
• Ensure all useful functionality (including activating/deactivating) can be done with keybindings that are commonly available - i.e., don't rely on Button2,3,4,5 as they may not be available with touchpads or some mice.

Outstanding Issues

• Some driver bugs such as an Nvidia resource exhaustion bug (black windows when too many are present with TFP) need to be evaluated if they are not fixed. A whitelist of cards needs to be created (or a blacklist) for enabling/disabling Composite. A wrapper needs to be written which does system checks, etc.