Unity? (was: What's your favorite distro, and why?)

Olli Ries olli at olli-ries.com
Wed Dec 11 16:49:31 MST 2013


On Wed, Dec 11, 2013 at 4:23 PM, Olli Ries <olli at olli-ries.com> wrote:

> On Sat, Dec 7, 2013 at 9:06 AM, Michael Torrie <torriem at gmail.com> wrote:
>
>> On 12/07/2013 07:30 AM, Olli Ries wrote:
>> > On Sat, Dec 7, 2013 at 1:18 AM, Dan Egli <ddavidegli at gmail.com> wrote:
>> >
>> >> On December 5, 2013, Michael Torrie wrote:
>>
>
[...]


>  You missed the biggest and most controversial change: Mir.
>>
>
> Mir does have nothing to do with Unity
>
>
>> Unity isn't all bad by any means.  I think part of the problem is
>> Canonical's decisions to go their own way in terms of Desktop
>> technologies (Mir, for example), instead of working with the community.
>
>
> well, there is a lot to be said (and it was back in March when Mir came ou
>

[major use of non-ML-compliant curse words for my touchpad deleted]

trying to restart...

well, there is a lot to be said (and it was said back in March when Mir
came out)

unlike Wayland or other other projects out there, Mir does not try to be a
general purpose system but is focusing on accomodating the requirements of
the guys sponsoring it, without limiting others to adopt it though. I
understand that one can argue that this is not the best way to drive open
source adoption but, at the same time, who cares as long as customer #1
(Unity) is happy when there is no other "customer".

Now there has been some unfortunate back and forth between Mark and e.g.
the KDE community, arguing about whether KDE will adopt it or not. Mir as
an OSS project would like to see wide adoption of other consumers and is
supportive of that. However, Mir tries to avoid an large extension
framework which will be X all over again (regarding incompatible
extensions, that are version incompatible etc).

The key argument why Mir exists today is simply that Wayland is only a
protocol definition, iow you need an actual implementation of it. This is
provided by Weston, which was referred to as testbed implementation
earlier. Moving that ever growing code base to something that is
"production ready" for an Ubuntu LTS release would have required a similar
amount of work than doing it "right" (for your very own definition of
"right") from scratch (think of TDD, CI, etc). Take all this and add a code
base on top that tries to cater to everybody & his dog's Desktop
Environment needs and you find yourself quickly asking as to why you would
go through all that when it can be done with a narrowed focus in less time
at higher velocity.
Writing a display server (i.e. a system that provides buffers that are
being rendered in, handles input and output) is not rocket science and
seeing Mir shipped on Ubuntu Touch should prove it. Other projects have
taken more time to get to a similar state and you would guess that most of
the time was spend on catering to many consumers' needs.

Mir is just another convenient addition to anti-Canonical arguments, I
don't remember seeing such an outcry when Google did Surface flinger (or
Android as a whole for that matter of not integrating well with the wider
OSS ecosystem).

In the (sometimes forgotten) OSS spirit of "each to his own" and general
software freedom, Canonical does not force anyone outside the Ubuntu
ecosystem to use Mir (where force is a strong word, the average Joe
UbuntuUser does not care if X/Wayland/Mir or something else is putting
pixels to his screen) and is stipulating Mir adoption simply by delivering
a (display server) system that rocks. Those that can see past the FUD of
CLA & not being a good OSS member & whatever else are giving Mir a rather
objective review and these end up being mostly positive.

my 0.02$
Olli


More information about the PLUG mailing list