A project blog would be amazing

[mdt]

@Caliga

But apps are important, too. If not for yourself, then for people around you that you would like to get excited about it. And I hope we, the backers, get involved in both. Like “this is our approach to power management, what do you guys think?” and also “We can only do the most vital stuff, but if somebody can work on an Alarm/Calendar/Notes/Calculator app, please let us know”

you touch the topic quite well: it’s an architectural question. be aware that apps where invented to have a sell-able portion that is detected as such by a non-technical because of its appearance as an icon. but it involves a special structure as you point out: you integrate gui and power managment, alarm and calendar. if you take a look at current apps you see a incredible bad design of a mix of different level of abstraction. i would like to prevent that.

i would like to have a design where function (alarm managment, power managment) and gui are strictly and cleanly separated. that introduces problems for marketing because nobody would pay 5$ for something that is not even visible in his launcher. but are not forced to follow a marketing architecture.

@Caliga

Out of interest, in your OpenMoko picture, what would be basement and roof? My current understanding would be that we have a pretty solid basement, from the kernel that should already support most/all components that shall be used up to the desktop environment.

i would say the kernel is the cellar, the basement is user space daemons, the roof is the gui…

the kernel may drive network devices, input devices and such. as you say: phonecall handling (and all modem handling) may be handled by a daemon. even network configuration is user-space and not kernel. exactly network handling is very, very specific on a mobile phone (failure is expected). you have to hand over between all different devices with no interruption for the user. dhcp, dns, routing, … NM cant do that reliable as of today. developer have to rethink the way they implement stuff if failure is the normal case. and the story goes on for higher-level stuff like contacts or even media. you /need/ a central contacts database to gain an integration between phone, email, chat and on.

so i would like to have a solution for the topic with a clear api that can simply be used by a gui program.

@Caliga

Of course I know, there is still A LOT that needs to be thought about. Will there be a daemon that handles incoming calls and starts an agent in the current user session? Power management etc. That could also be considered basement. And as I am a developer for a Linux application myself, I know how much difference there is between “basically works” and “works pretty nicely”. I hope it all goes well…

you name it: there is alot to do on top of what a desktop environment could provide. i think the road coming from a desktop environment like gnome down to a mobile phone is long and you wont make it in one and a half year (other projects already proofed that ).

instead if one provides the groundwork and define common APIs its much simpler for the community to provide GUIs on top. and a look into other solutions isn’t bad. android has alot nice design decisions (and bad ones as well) that can help to understand what to do…