The backend will likely use Postfix as MTA and dovecot (or other more lightweight solution) as MUA. Haven't settled on a final configuration yet. The magic will happen with the frontend and the utilities that oversee its operation, of course, and those will be custom developed.

I'm also working on a system for optional caching (encrypted, of course) which would help if your arkOS node is knocked offline, as to not screw up mail processing locally or for your communication partners. But that is still in rough planning.

Sorry if I missed this somewhere but how does arkOS intend to handle delivery issues? Many consumer ISPs make it very difficult for an end user to have good mail delivery rates, and especially so to freemail providers like Google, Hotmail(MSN), Yahoo, etc.

Without getting into it too much (i could write a novel :p ), the caching system will have a role to play in this. There are many different ways to solve this problem, and rest assured that creating a competent and stable mail system (and all that entails) is top priority.

As a lot of the system already uses python, looks like lamson might be a good fit:

  http://lamsonproject.org/

I'd love to hear more about why a whole new distribution, rather than some management tools on top of something like raspbian?

I understand that targeting a single platform is easier -- for an OS distribution -- but it sounds like what you actually want is an application server on top of a base os -- one that allows for a few standard tasks (email, web etc).

Why reinvent all the other parts of the stack?

Thanks for the link, I'll check it out! As far as it being a new distro is concerned, my goal was to make it as simple and powerful as possible. Simplicity means being able to have a clean download and install process, prepackaged for those who might be intimidated by having to install Linux on their own. Also being able to control package versioning and selection is critical for ensuring the dashboard remains as stable as possible. Powerful because it allows for many cool utilities to be installed, configured and ready to go with minimal intervention. For example, one of the existing utilities allows the RPI to cache log writes in RAM, reducing SD card wear which is critical when using the Pi. So there are many ways to dress up a system like arkOS. I acknowledge that making it its own distro is a more thorough solution than it would be otherwise, but at the end of the day I feel it is the one that allows for the best end user experience.

It still sounds like you're doing a lot of extra work for little gain vs making a Debian remix or something. I realize writing your own can be more fun, and seem easier, than establishing detailed knowledge of an existing system -- but all of the things you mention could be done quite easily starting from a working Debian distro.

You could fork into a complete distro, like Ubuntu, or just choose a few (core) packages and add some patches.

Either way, I wish your project the best of luck -- any project that tries to revert the asymmetric split of the Internet in "end-users"/consumers and "services" is welcome in my book :-)