[prev in list] [next in list] [prev in thread] [next in thread] 

List:       meego-dev
Subject:    Re: [MeeGo-dev] MeeGo Reconstructed - a plan of action and
From:       Jeremiah Foster <jeremiah.foster () pelagicore ! com>
Date:       2011-10-05 8:55:39
Message-ID: CADVDRBjMwU+cfWXUXazLD2rbeaH7fJVfn6jzq+xqagZv-50Cgw () mail ! gmail ! com
[Download RAW message or body]

On Wed, Oct 5, 2011 at 9:39 AM, Samuel Stirtzel
<s.stirtzel@googlemail.com> wrote:
>
> 2011/10/4 Jeremiah Foster <jeremiah.foster@pelagicore.com>:

> >> Can a mobile
> >> segment distro like MeeGo be really compared with a desktop segment
> >> distro like e.g. embedded Ubuntu? (This is not relative to your
> >> message but a general rhetorical question.)
> >
> > I don't understand what you mean by "segment" here. And I also don't
> > understand what you're referring to with "embedded Ubuntu."
>
> What I meant is that a embedded desktop OS targets another audience
> than the embedded mobile OS (or in other terms a different market
> segment).
>
> Well I'm sorry for using the term embedded Ubuntu, I've assumed that
> others refer to projects like Linaro-Ubuntu [1], the TI-OMAP Ubuntu
> [2] and in general ARM Ubuntu as "embedded Ubuntu" too.

No reason to apologize! I was just not certain what you were referring to. :)

> >>
> >> > Before it was just big companies that could create their own Linux
> >> > distros
> >> > (before that everyone had their bespoke UNIX distro) nowadays
> >> > fragmentation
> >> > is brought to you by every Tom, Dick and Harry with an OBS login.
> >> > I've been down the fragmentation road before. It always ends with
> >> > retracing
> >> > your path back to the main highway.
> >>
> >> There seems to be much standardization work going on in the Yocto
> >> Project / OpenEmbedded Core (see [3], also Carsten already mentioned
> >> the Yocto Project in context of governance), anyone evaluated it?
> >
> > Yocto is very interesting indeed, as is OpenEmbedded, though the claim is
> > that Yocto is "open embedded done right." But for the purposes of a distro I
> > don't think Yocto is the silver bullet people are looking for. Firstly, it
> > seems focused on Intel Atom BSPs and overall seems designed to help in board
> > bring-up. Yes you can create a complete distro, but like a misused OBS
> > repository, creating your own complete distro is not a good idea. Unless of
> > course yours is THE ONE.
>
> Your statement about Yocto is right, but OpenEmbedded isn't
> OpenEmbedded anymore, alot has changed since this statement was true.
>
> OpenEmbedded Core (the new OpenEmbedded) and the Yocto Project merged
> their efforts and created an unified code base

I tried to point that out by saying "Yocto is 'open embedded done
right.'" But I guess I was unclear. From what I understand bitbake is
still at the heart of both these projects and that is a pretty
interesting piece of technology.

>
> (see [3] and [4]), also
> OpenEmbedded Core + Yocto supports a larger audience (see [5] for the
> approved list of BSP layers).
>
> In OSS systems there should not be a THE ONE distribution, for end
> users this is out of the question, they might not want to create their
> own distro, but (IMHO) developers should not be restricted in this
> direction.

But that is precisely the point. To serve end users you need a
consistent, easy to use platform, with a large ecosystem of developers
to create new applications. This is one of the reasons why Debian is
useful: its foundation is that it puts users first. From the Debian
Social Contract; "We will be guided by the needs of our users and the
free software community."

Yocto puts developers first which is great - but it doesn't make for a
consistent, easy to use platform, with a large ecosystem. It creates
different chunks which are little Linux stacks which one places over
the "layers" which seem to be like Linaro Hardware Packs, i.e. BSPs
and drivers. These chunks function as a Linux distro, you can add
packages and UIs and all kinds of amazing stuff, but you have no
community. You have no ecosystem, you don't scale.

> >>
> >> The Yocto Project / OpenEmbedded was discussed before in the IVI
> >> mailing list (see [4]), but it lacks any technical explanations and
> >> arguments why it cannot be used / or get adapted, also OpenEmbedded
> >> progressed into the new OpenEmbedded Core Project (the next release is
> >> just one step away).
> >
> > https://wiki.yoctoproject.org/wiki/FAQ
> > OpenEmbedded is widely used in commercial embedded systems. Those systems
> > tend to not be open source systems and end up costing lots of money.
> > Regards,
> > Jeremiah
>
> Sorry I don't get your point here, no offense but if I would say:
> "Debian is widely used in commercial desktop systems." Would your
> statement be the same?

No offense taken. I was purposely vague because I don't want to
publicly deride commercial Linux companies and other hardware
companies that use OE and Yocto.

The difference for me is that commercial Linux companies see Linux as
a technology. In my mind, that is a misunderstanding. Linux is a
kernel, combined with a userland (Android, GNU, etc.) and a license
group. This combination is where the magic occurs because it enforces
useful changes back into GNU/Linux distros and encourages a thriving
ecosystem. This thriving ecosystem is what makes Linux run on
everything from MIPS, to Xeon, to ARM M5, to A11, from old desktops to
the world's fastest computers, from phones to satellites.

> If you care about the patches, i see quite an amount of users sending
> patches back to OpenEmbedded Core and Yocto.

There is no doubt that OE and Yocto are open. There is no doubt that
their communities are growing and contributing back. My doubts are
around the fragmentation Yocto creates. This is why I am skeptical
about Mer and even Tizen. Do we really need another build system? Do
we really need another distro? If these things don't achieve critical
mass to be sustainable, have we wasted the talents of important
software developers while they worked on these projects?

I don't see compelling answers to those questions. But maybe that's just me.

Regards,

Jeremiah
_______________________________________________
MeeGo-dev mailing list
MeeGo-dev@meego.com
http://lists.meego.com/listinfo/meego-dev
http://wiki.meego.com/Mailing_list_guidelines


[prev in list] [next in list] [prev in thread] [next in thread]