>>1931(cont. last)
I probably missed some of your comments. Sorry it's late.
My last comment for now. I envision the eventual set-up to look something like this;
- Base system
This will be the "core" of the OS. It will be GCC+the GNU tools. Provides a standard POSIX implementation and basic hardware support for all AMD64 systems. Should be able to run all Linux and *BSD software. Small. Minimal. Nothing goes in here unless absolutely needed even by the most minimal of installations.
- Standard desktop
This is the next level up. Provides advanced/modern graphics (e.g. anything that's not the frame buffer) and standard suite of libraries applications and tools to run the basic desktop(s). The "minimal" installation for people that want more than just vim, awk, grep and other CLI tools. Things like ffmpeg, wine and qemu go here as well.
On top of this you could add most anything you want of course. I won't go into details now but you'll be able to access most everything from repos like the AUR and GURU. Although sometimes not directly unless you want to do additional set-up (see: Bedrock Linux). Through simple config files/changes in GUI you can substitute any application you want for another. You can also have as many versions/copies of each application installed as you want and they will play nice together.
The long term goal: The system should become smaller and simpler as the years go on. There are _many_ things about the current set-up I'm running that I _really_ dislike. But they've been retained because they're a PITA to hack around and break some common user configurations at the moment. An example would be dbus. Something I wish to remove all together and can on my personal install. However, some common things like Steam rely upon it. I'm also not opposed to the idea of IPC daemon itself I just dislike the dbus implementation of the concept. Solution? Make a better IPC daemon. I've been helping someone work on one for awhile now. But it isn't ready or stable. Once it's stable I'll switch over to it then provide a way for it to emulate dbus. In time. Hopefully others will prefer ours to dbus and switch over to. There are many things like that in the current base system due to the state of Linux development world.
I'll try to remember to come back and post updates. Thanks for your attention. I wish I had more time to work on this because I would have gotten it out a couple of years ago now if I did. But it kind of turned out to be a good thing. Since some improvements happened in the distro I'm basing it on recently that were good for once. So now is a preferable time to fork compared to years ago. On the other hand. Another project I was pulling a lot of code in from folded recently. Leaving me to maintain that stuff myself.
I am hopeful we will be able to make improvements to the overall *nix ecosystem while avoiding the ideological debates that mostly serve to prevent any real discussion or improvements from happening. But I know the attack dogs will come out in full force when I do eventually publish this. The best thing we can do is ignore them, patch/hack around their nonsense when needed, fork when required and provide a better overall user experience than they do. The last one isn't that hard. It just takes the will to care about the end users and not assume they're brain dead. In other words: We just have to be nice to them and write good documentation.
Also in time this should be able to run on many non-x86_64 archs. It already does. But starting out AMD64 is the only platform that will have published binaries that don't require building everything from source. Speaking of that; One requirement for any package within the repos will be that it builds from source. You will be able to mix and match bin and from-source packages. Also unlike most distros (like Debian) all dev tools will be installed by default assuming something built with them exists in the system.