Some thoughts about next steps.
Turning these first few steps into a real Debian port is going to take a serious amount of work. I've read through a number of the early email threads on the debian-arm email list involving the armhf port and it took an enormous effort to get the that project started. Much of that effort continues today as armhf problems are still being resolved in various packages. Of course, an RPi flavored armhf port can leverage 99% of that effort, but even if it is just a task of recompiling and distributing the packages, there are 10,000's of Debian packages that need to built and managed.
First of all, if this is to be done, I would really like to see this be done right. This project has a lot less appeal to me if the end result is only a minor subset of the Debian packages being made available that are just used by a handful of RPi hackers. My own personal goal would be to make an RPi optimized Debian port as easy (of course that's a relative term) to use on the RPi as it is on any of the other platforms supported by Debian. Any end user should be able to start with a basic Debian boot image on an SD card and then customize their system from the myriad of packages supported by Debian. They should know that what they download will function as advertised and will run as efficiently as possible on the RPi. If that can be achieved, I believe Debian would indeed become the premiere Linux distribution on the RPi and attract thousands of new users into using Linux in general, and Debian in particular.
Doing things right includes honoring all Debian licensing terms. Much of it involving GPL. It is particularly important that the project begin on the right footing in the this regards if we are going to attract the support of the Debian developers and community which is important for success. This means that in addition to making the binary packages available, the source code also needs to be managed in such a way that it is available for download as well for a term of 3 years from the binaries being available. I've already been told by people associated with Debian that simply pointing URLs at the source code for Debian packages is insufficient to satisfy the GPL licensing terms -- the source needs to be located with the binaries and equally accessible. I mention this as it does increase the logistics of making packages available. Also, how to handle the proprietary packages that would come from closed binaries from Broadcom and the RPi foundation would need to be resolved as well in a way consistent with Debian practices.
BTW, honoring the licensing terms is what is going to keep me from making available for now the binary packages I've already compiled. I simply don't have the time or resources to honor, to the letter, the licensing terms with regards to source code. However, I'll look into what I can do about this over the next week. Although it may be more convenient, ignoring the terms now would not be setting this project on the right foot for long term success.
At this point, I'll need to get some Debian mentors. These mentors will be people already established in the Debian hierarchy with upload privileges and they will have tribal knowledge of how to get things done "the Debian way". Hopefully, because RPi is already so well known, finding mentors that will support a Debian armhf flavor for the RPi won't be a hard task. The harder task will be convincing Debian mentors that I and other people interested in this project are serious and have the skills and motivation to see to the project through to success. I'll start putting out some emails on this and see how far I get.
Another aspect to be considered is what hardware infrastructure will be required to compile the 10,000's of Debian packages and keep the packages refreshed as new versions are made available. I simply don't yet know what would be involved with using existing Debian resources to build such packages and what might be involved with applying to use the resources -- this is where a mentor could help answer a lot of questions. Looking over how other Debian ports began, it seems they usually start with hardware resources outside of Debian and as the port achieves success then Debian begins to consider how to fold the process of building packages for the port into the Debian build infrastructure. If this is the case, this would involve purchasing and managing a build cluster of at least a half dozen systems to get the port underway. This is usually done with the sponsorship of a company or university, but I don't see that happening here unless someone knows of a company willing to help sponsor a Debian port with financial and other resources. As long as the costs can be kept reasonable, I can probably cover the expense of such a cluster (although I'll have to pass this by my wife .
So, there are my early thoughts on what the next steps will be to make an RPi flavored Debian armhf port a reality. More immediately, over the next week I'm going to investigate automating the Debian package build process -- most likely using sbuild. This will undoubtedly involve hand building dozens of additional packages to install the sbuild. However, once done, it should be easy to replicate and get other build systems set up. Also, I'll need to start building the packages without cutting some corners such as skipping digital signatures, lintian processing and such. I also have lingering issues with the compilers and with the debootstrap process that need to be resolved.
I'll keep posting in this thread the status of progress made against the items discussed above.