> It may not be hidden but, after five years, there is not a single product or distro that uses it or even has a timeline for when they will use it (correct me if I am wrong). Ubuntu wants to put something out in the next few months, and unify all their target devices by next year.
This is probably the biggest issue that has created Mir. There have been a few links to a post by Shuttleworth from two years ago which discusses a presumed shift to Wayland that indicated Canonical wanted to integrate it in the next 6mo cycle. That obviously didn't happen and Wayland didn't seem to stabilize on the schedule needed for Ubuntu integration. Maybe there just isn't enough resources behind Wayland, did Canonical have developers working on it or did they expect that others would do all the work without their resources? It seems that they would have been hot to integrate it if it was ready a year ago, which would have probably prevented this whole issue.
Ubuntu unveils its next-generation shell and display server
Posted Mar 7, 2013 6:59 UTC (Thu) by airlied (subscriber, #9104)
[Link]
Canonical have never provided any resources to the wayland project. They have waited for some other distro to do the heavy lifting, you have to also remember the wayland devs have not just been working on wayland for the past 5 years.
Most of krh's time has been spent getting the mesa gbm/egl layers to a place where wayland is actually possible, and open source stack supports it.
If someone had come along a year ago with a direction for wayland and 4-5 developers there would have been very little to stop them taking the project in whatever direction they wished. You'd at least talk to the upstream project first, to see what benefit you could gain from it.
There's a pretty good chance Fedora 20/21 will see some movement in waylands direction. But if Canonical had put the effort into wayland instead of Mir we might have had a reason to help out earlier and assign resources.
Ubuntu unveils its next-generation shell and display server
Posted Mar 7, 2013 14:02 UTC (Thu) by dbnichol (subscriber, #39622)
[Link]
Writing a display server from scratch somehow accelerates their product? That doesn't make any sense at all. Not to mention that they're now doing this completely on their own now and will receive no help from the open source graphics veterans working on Wayland now like Kristian and Daniel Stone. People that have been doing this for years and have a pretty good idea of what a display server (X) should and shouldn't be doing.
If time to market was all they cared about, why not use X? Or why not just fork Wayland/Weston since that code already exists? Or just use SurfaceFlinger? I don't think time to market makes sense as the driver for Mir. It seems a lot more about control.
Ubuntu unveils its next-generation shell and display server
Posted Mar 7, 2013 17:05 UTC (Thu) by raven667 (subscriber, #5198)
[Link]
> Writing a display server from scratch somehow accelerates their product?
Not having to integrate or coordinate with anyone else accelerates development, just as Fred Brooks 8-)
> People that have been doing this for years and have a pretty good idea of what a display server (X) should and shouldn't be doing.
They might be able to get something lightweight that works 90% for their one use case without worrying too much about the corner cases, it's the last 10% to make it robust and complete where the difficult work is. This reminds me of mjg59's criticism of LightDM, by the time you've solved the real-world problems and discovered all the underlying requirements your "lightweight" system has just as much complexity as the "crufty" thing it was meant to replace.
So I think they can ship something that pretty much works, but it'll take years to get to the same place that Wayland is today, just like Wayland took years to build, and then they'll have another big of private infrastructure to maintain that takes away time from actually adding value to what they are selling.
Ubuntu unveils its next-generation shell and display server
Posted Mar 8, 2013 13:56 UTC (Fri) by robclark (subscriber, #74945)
[Link]
> Not having to integrate or coordinate with anyone else accelerates development, just as Fred Brooks 8-)
well, maybe for whipping together a prototype.. for getting something that handles all cases (multi-display, hotplug, various different apps, etc) it is not. In the end we just end up with something that works ok in simple cases but is not as mature/robust as wayland, and everyone loses.