Two days ago we reported on what the graphics stack should look like for Ubuntu 11.04
in terms of its X.Org Server, Mesa / Gallium3D, and the open-source graphics driver versions to be deployed in this next Linux operating system release codenamed the Natty Narwhal
. This though wasn't the only X-related discussion to take place at the Ubuntu 11.04 developer summit
in Orlando this week, but there were other related topics discussed such as KMS configuration / quirk handling, the multi-monitor experience on the Ubuntu desktop, and multi-touch support. There were also talks aimed at Linaro
/ embedded Ubuntu on ARM platforms with regards to embedded GPU drivers and OpenGL ES support.
While the quality of the notes vary from session to session, information from these talks at the UDS Orlando event can be found by connecting to gobby.ubuntu.com
from a Gobby 0.5 client for collaborative document editing / sharing.
In regards to X.Org configurations, Ubuntu developers are deciding how to handle the small percentage of desktop users where the out-of-the-box monitor experience is ideal with either the monitor(s) being set to the wrong resolution, no display at all, or similar problems. These problems are mostly attributed to monitor vendors that have models where the EDID (Extended Display Identification Data) is incorrect or even not supplied at all. Back when the open-source drivers relied on user-space mode-setting, affected users could use a variety of options within the xorg.conf
to override values and handle quirks within the server. However, with kernel mode-setting that support is out the window.
To address these issues, Ubuntu developers will be investigating ways to make user-supplied EDID being writable at the kernel level, moving Intel driver power configurations and such to a sysfs interface, extending the GNOME Monitor Applet to support extra RandR modes, summarizing available quirks, and a proposal for a device tree node specifying the graphics quirk.
In terms of the Ubuntu multi-monitor experience, Ubuntu developers are trying to decide how to handle different use-cases. For instance, if a projector is connected to a laptop, should the output be automatically be enabled and set to clone mode for a presentation? What happens if no monitors are connected? What was suggested during the session was to default to clone mode when connecting a second monitor that defaults to the smaller head's resolution (though this is a limitation anyways for many X.Org drivers), allowing the user to specify monitor preferences, and a configurable hot-key for launching the display chooser.
There are also multi-monitor problems to be explored upon monitor disconnection, dialogs being displayed on secondary displays, Ubuntu's Unity
defaulting to the left-most monitor to be primary, and no display heads being connected to the GPU will cause the xorg-server initialization to fail.
For those connecting to the Ubuntu Gobby server to read all of the notes in full on these discussions, some of the documents worth checking out include: multimedia-desktop-n-xorg-multi-monitor-experience (the multi-monitor support), multimedia-dx-n-xorg-and-mt (state of X and Multi-Touch with X Input 2.1, UTouch, etc), multimedia-arm-n-gles-in-ubuntu (OpenGL ES for Ubuntu on ARM), and multimedia-desktop-n-xorg-general-planning (the general X.Org talk we commented on two days ago).