[Openicc] GSoC 2013 preparations

[Kai-Uwe Behrmann]

Am 19.03.2013 08:53, schrieb Richard Hughes:
> On 19 March 2013 07:23, Kai-Uwe Behrmann <ku.b at gmx.de> wrote:
>> After you suggested ucmm several times, we repeatedly showed that we build
>> upon it's basics, last but not least during the last CM hackfest in Brno. We
>> rely on that.
>
> Right, but I was talking about ucmm, not the extended version you/we worked on.

Ok

>> We successfully collaborate through the above linked JSON format in various
>> projects. JSON support is differently implemented and deployed by those open
>> source CMS projects. That all makes JSON already a part of our history. That
>> all is only a small step away from ucmm.
>
> It still doesn't solve the atomicity problem, or the many-writers cache problem.

Can you elaborate on this problem. I do not see here that CMS 
configurations are frequently written. If they need persitent 
frequently rewriting, then there is something wrong with the underlying 
architacture. Just to be clear, we talk here about device configurations 
and some few settings.

>> The need for a external library code has proven to be a show stopper for
>> collaboration. That is the lesson learned by colord joining the Linux
>> platform IMO.
>
> I disagree. The number of projects with a colord dep are growing every
> day. See http://qa.debian.org/popcon.php?package=colord for some real
> world numbers.

... plain installation statistics. They tell not much about dependencies.

>> Comming now with a very new SQlite scheme would impose a
>> useless burden on already collaborating projects.
>
> Well, over 60 different packages on my box already depend on sqlite,
> so I don't really buy that argument, sorry.

My comment was coined towards pre existing specs. And those rely on 
JSON. SQlite is in that context just a new thing without much novelity 
or improved flexibility.

kind regards
Kai-Uwe