Introducing The X.Org Gesture Extension
Phoronix: Introducing The X.Org Gesture Extension
Earlier this morning Canonical had announced the UTouch Framework, which is their multi-touch framework to be formally introduced with Ubuntu 10.10. Canonical developers crafted up their own multi-touch solution and even their own gesture language for Ubuntu, rather than leveraging any similar free software projects, but -- to some surprise -- it turns out they are now going to try to engage with upstream developers to at least have a formalized extension to the X.Org Server for gestures...
woo! yay canonical for going upstream with something this important!
maybe now i can use my wacom bamboo graphics tablet correctly?
obvious question: does Android have an open-source multi-touch protocol/component that could be ported to X?
Originally Posted by phoronix
from canonical's press release it sounds like they have been collaborating with upstreams all along. it even credits all the people whos work they have leveraged.
They just got AppArmor into the kernel too - hopefully this will help address some of the criticism about lack of support for upstream which has been levelled at Canonical recently.
Originally Posted by portets43
Which is kinda stupid, really. They are involved in:
Originally Posted by krazy
- system boot, with Upstart and initramfs-tools, now adopted by other distributions
- package management, with dpkg, gnome-app-install, update-manager, apt, Synaptic, gdebi, which all goes back into Debian
- installation, with debian-installer and Ubiquity, which is used by Debian and derivatives
- crash reporting, with Apport, used in Fedora
- Linux Terminal Server Project, developed by all distributions, led by Ubuntu for 2.5 years
- QA for Xrandr
- KDE, with System Settings, a printer applet and power manager
- employing some GNOME developers
- bugfixes (patches.ubuntu.com), pushed upstream
- hosting svn.gnome.org, l10n.gnome.org, Inkscape, Miro, etc
- Uncomplicated Firewall
- Uncomplicated VM Builder
- AppArmor, pushing it into the kernel
- infrastructure, with IdeaTorrent, Launchpad, GNU Bazaar
- server software, with PowerNap and Byobu (extension of GNU Screen)
And now with the proposed multitouch X11 extension, people are saying that they 'finally' contributed...
"Chase Douglas of Canonical has just announced the X Gesture Extension to the X.Org development community and he hopes this will become part of the X Input 2.1 extension. The X Gesture Extension is to work hand-in-hand with the recently-drafted X.Org Multi-Touch Protocol Specification that was written by input-expert Peter Hutterer. "
This is definitely an improvement on the past, and worth congratulating Caonical for. But there's still a problem:
"The X Gesture Extension will be found in Ubuntu 10.10, which is using X.Org Server 1.9 but with Canonical's patches. The Multi-Touch and Gesture protocols for X.Org are planned for X Input 2.1, which could come with X.Org Server 1.10 due out early next year."
Announcing that you're going to include it in Ubuntu 10.10 - presumably regardless of whether upstream decides to adopt it or not - is rather holding a gun to the head of the upstream consideration of the proposal. Again, this is still better than previous efforts and that's great, but the best way to do this is to draw up your plans, submit them to upstream at that point, get them adopted, work on the code in upstream, and *then* pull it into Ubuntu once it's released as part of a new X.org.
okay, so now I read the thread (which I really should have done before, bad Adam) I see it's not too terrible; they're releasing 10.10 with an extra abstraction layer on top of the draft version of the protocol and they'll be building whatever multitouch stuff they need on top of that abstraction layer, so if the underlying protocol changes with upstream adoption and they take the upstream version for 11.04, they can update the abstraction layer for the underlying protocol changes and the code written against the abstraction layer for 10.10 will still work.
Originally Posted by AdamW
It's somewhat over-complex, but it seems like a reasonable way to compromise on upstreaming versus what I suspect is some kind of OEM need to have viable multitouch support in 10.10. So that's really not bad at all. Decent job.
Please stop this list, it looks like you are more doing a salespeach for Ubuntu then actually listing where they have contributed upstream (remember being upstream is not the same as contributing upstream), and not being upstream they are pretty good at sometimes. I think one of the reasons for Fedora to look for other alternatives then upstart was the lack of interests from the maintainers @ Canonical in older versions even if they existed in still supported versions of ubuntu.
Originally Posted by Remco
Yes Canonical does much to give back to the community, but for example the openssl fiasco some years back shows how bad they are at it sometimes (and yes Debian should have some flogging for this too) where they instead of bringing a patch upstream and getting a comment on why the patch is bad they just incorporate it and leaves their users whith badly broken keys securitywise.
Welcome to the real world. That's how successful standards evolve: you implement something, you prove its worth and then propose to extract a standard from your work. Think how the ARB evolves OpenGL: an IHV produces some vendor-specific extension ("MESA", "ATI", "NV", etc), then a second one starts supporting it (it gets renamed to "EXT"), then a third (it gets renamed to "ARB"), then it moves into the OpenGL core and everyone is expected to implement it in order to claim conformance.
Originally Posted by AdamW
The opposite process, produce a standard out of thin air and then expect people to implement it, almost never works.
Tags for this Thread