I haven't noticed this thread until now, so I'll express my opinion of what Puppy 6 should look like.
I see it as the direct descendant of Puppy 5.x, but more organized and efficient, since the core Puppy developers have gained lots of experience with Woof and know more about the distributions it supports (most notably, Ubuntu, Debian and Slackware). I also think it's time to start thinking about making Puppy more "serious" in terms of infrastructure and the community structure; I think we should implement some automated package building solution (to make Puppy more source-based) and form a team of lead developers, which will coordinate the work of small, task-specific teams, such as a "kernel team", an artwork team, etc'.
However, words are words, time to talk about code.
At the moment I'm working on 4 components that form the infrastructure needed for a modern distribution:
- A competitor to Woof - I implemented something very similar to Woof, but more modular, much faster (15 mins vs. Woof's 5 hours, on my netbook, from the package processing part to the bootable image) and smarter. The main features are good design, efficient code, zero legacy code and generic nature (e.g less hard dependencies on specific packages). It's fully compatible with Woof's configuration format, so I was able to build a Slacko with it; the result was quite crude (i.e it was way bigger), but RAM usage was about 10 MB lower.
It's ready and works great.
- A package building server - an automated, complete and sanitized solution for building packages. This server contains well-tested build scripts for all applications shipped with the distribution; the rest is taken from another distribution. This means we have good balance between size and ease of development: the base packages are provided by some "upstream" source, while the part of the operating system users care about is totally customizable. This server makes sure the packages are built in a clean environment using jails and virtualization, so development is easier and more future-proof.
I use it for several months and it's great.
- A distribution, all the surroundings - I wrote build scripts, a package manager and documentation for a proof-of-concept Puppy "fork", which also has its own branding. This use of the term "fork" is kinda silly, since it's an independent distribution, but it's aimed at showing the world what my vision of Puppy 6 looks like; I'm a realistic person - building a POC of something is always better than explaining what it is in words. All its applications are built using the package building server, while the core binary packages come from Slackware (or any other distribution, I just need to implement support for it in my Woof implementation).
- A file system skeleton and init scripts. The package manager is ready, as I mentioned earlier, so I just need to write new init scripts. My experience with building small distributions should make it easy.
It's kinda funny that I find myself working on all these, since these are things I've already done in my past, but separately. I built my concept distribution (Calf GNU/Linux), which had its own skeleton, init scripts, package manager and binary packages, but lacked a proper building system.
Now I decided to go for it and build all 4 components, so I can see how far it gets. Unlike Saluki and other innovative Puppy initiatives, I'm starting from scratch - everything is fresh and clean,
as the next Puppy generation should be. I'm not going to use Woof or Puppy's existing tools, because the best revolutions aren't transformations or natural evolution, but simple innovations that make things better.
2c