challenges of distro switching

[Daniel Drake]

On 5 June 2010 12:14, Tiago Marques <tiagomnm at gmail.com> wrote:
> I built OLPC's kernel in Gentoo with no problems, which helped me a lot to
> start. The rest works great.

Perhaps we aren't talking about the same things. I'm referring to how
we can build an image which would be suitable for OLPC 1:1 deployments
at large scale, without introducing huge churn for existing
deployments, without demanding a significant additional amount of work
both now and ongoing.

I suspect you're only working with what is important for you. (i.e.
you aren't using antitheft, you aren't pushing OS updates
automatically to yourself, you aren't using the customization stick,
you aren't customizing the Browse homepage, ...)

> Can you be more specific of what more would break? I can think of
> olpc-update but not much else unfortunately.

Antitheft
Activation

A mountain of packages and their corresponding effects on the system,
to think of a few: olpc-update, olpc-contents, ds-backup, bitfrost,
dracut-modules-olpc, library, kbdshim, powred, runin, olpc-utils,
switch-desktop, bootanim
Several of those packages depend on Fedora-specific things that would
need to be adapted.

Various important bits of Fedora that we depend on - like their
memory-backed rwtab file system.

New build system would be needed, replacing all functionality of olpc-os-builder

It also invalidates many processes that myself and others have been
training in the field, such as usage of rpm/yum, how to write spec
files, etc.

It would result in a range of different package versions being used,
compiled against different dependencies, invalidating a lot of
testing, creating divergance from XO-1.5.

It means everyone working with OLPC at a suitably low level has to
learn a new distro.

> That's true but you're again limited to not having security fix or updates
> available after 12 months(?), which from what I've seen is around the time
> the image is ready for deployment.
> I have no idea of the work involved in that, aside from what I read here.
> From what I've read it could be worth it but it was just a suggestion.

Generally not important for field deployments. In the cases where it
is important, OLPC already has a mechanism (push a new RPM, maybe
generated manually by OLPC, and do a new release).

I'm not discounting your suggestion to switch to another distro. Feel
free to steam ahead with your efforts.

I was solely talking about a simple way to keep the existing base
available to XO-1 deployments. The thread that you hijacked was about
Fedora.

Daniel