New XO-1.5 10.2.0 build 119

[Jon Nettleton]

On Mon, Apr 12, 2010 at 2:04 PM, Peter Robinson <pbrobinson at gmail.com> wrote:
> On Mon, Apr 12, 2010 at 10:04 PM, Mikus Grinbergs <mikus at bga.com> wrote:
>>> It will then be OK for a while until Sugar needs a newer version of a
>>> library
>>
>> I think there is an "elephant in the living room" to which too little
>> attention is being paid.  Consider Peru - they have a lot of XO-1
>> systems, on which it is unlikely that the newest version of Sugar will
>> ever be installed.  But Peru *might* want to add a Linux utility into
>> their systems -- for that they need access to a package repository with
>> the same-level software as the rest of the software they already have.
>>
>> What I am saying is that for many deployments, a "need for newer
>> libraries" will probably never exist -- but a "need for older libraries"
>>  is more likely to arise.
>
> I'm not saying its not the right direction and it might be almost time
> to move there but I've also seen hacked up versions of old Fedora in
> the past just to get the later version of some telepathy library, and
> Tomeu's work to bring sugar inline with the rest of telepathy and
> allow support of other IM interfaces in the next release might well be
> an example of this.
>
> Believe me I'll be one of the first poeple adding the sugar packages
> to EL-6 branches and testing them but its not necessarily the golden
> path that some people think. In the short term of the first 12 months
> it will be fine, 18 months to 2.5 years it won't seem as good.

I am not seeing where the 18 months to 2.5 years breakdown happens.
You have a base OS that is being supported for security updates by the
largest Linux vendor for the next 6 years.  For the small subset of
packages that you need a more recent version of, i.e. the kernel and
telepathy, you would just release in an OLPC yum repository.  This is
already being done for the kernel and other OLPC specific packages, so
really there is little overhead to this path.

We are getting to a point where the core components of linux on the
desktop are matured such that a core refresh really isn't needed every
6 months.  NetworkManager, PackageKit, PolicyKit provide a lot of what
the desktop needs from an administration point (Yes I didn't mention
Pulseaudio, maybe soon :-)).  For the first time in a long time I
really don't see any major changes in the near future that will really
be a roadblock to a stable distribution rolled out in the next year.

-Jon