You are not logged in.
Pages: 1
Hi All! So for the past few months I've been starting the beginning stages of making a new distro, initially I was going to use NetBSD's kernel, but because of hardware support I chose linux-libre, all of the source is contained in one bsd-style directory and build-system, it builds the entire GNU userland statically against musl-libc, nothing in the base system is dynamic, including the kernel, it's completely static, void of ancient drivers, and only supports loading external modules like vboxdrv, I have also patched it to reflect "GNU" in uname. All of the third-party packages will be built using arch linux and pkgsrc, so anything that needs shared libraries will be completely self-contained, the only dependency in the base system in the runtime loader which is now /sbin/rtdl instead of in the lib directory. I have Xorg working and some small programs, but right now, I need a working installer, and to start building all third party packages statically. Ill upload my changes to github tomorrow at some point, I just wanted to see if anyone is interested or has any questions about it?
Eventually ill replace linux kernel, with a BSD kernel, probably Net/Openbsd or Dragonfly, Im only waiting on decent kms support.
Initially I was going to port gnu to minix3, or try to do some work on hurd (sound,sata etc..) but I realized at this point in time, any microkernel based system would not get enough attention to be worth working on, but In the long run, I would like to use hurd, if it ever gets better hardware support, as that is the only REAL problem with it.
Offline
In my mind, this is a blog post bordering on solicitation; I had to read the post twice before really noticing the "any questions?" prompt, which isn't the most constructive way of starting a discussion. And since this is in no way related to Arch (how do you use Arch to build third-party packages from the pkgsrc collection?), and ultimately aims not to be Linux-related etirely it doesn't really belong in the "Community Contributions" forum.
Offline
ultimately aims not to be Linux-related etirely it doesn't really belong in the "Community Contributions" forum.
My issue with it is that there seems to be no code yet. Ideas are neat, but it's the execution that counts.
Is the a website, a github repo etc. we can have a look at?
Offline
Maybe check out http://starchlinux.org/, it seems to have a similar goal and already has some code.
Offline
stali also seems close to the description.
Offline
As nothing has been contributed so far, I have moved this to "GNU/Linux Discussion". Send me a message when there is something to share and I will move it back.
My Arch Linux Stuff • Forum Etiquette • Community Ethos - Arch is not for everyone
Offline
stali also seems close to the description.
It's also vaporware.
Offline
Subscribed. I look forward to seeing some code in the git repo!
I'm in the process of doing something similar although it's more of a HLFS project with a lot of revamping going on (same as you--static + musl libc and whatnot). So, I look forward to seeing how you do things so I can possibly learn what I can do or change in my own endeavours.
Offline
ANOKNUSA wrote:stali also seems close to the description.
It's also vaporware.
Indeed, but the basic ideas are still there as a jumping-off point. I imagine the trick would be actually getting something working faster than the software ecosystem can change; it's probably much more difficult in the Linux realm than the BSD realm, where core systems change slowly and deliberately in a unified manner. Just out of curiosity: It seems that your ultimate goal is to use a *BSD kernel; is Linux just a starting point then, given the extant tools and support, or are you working more toward a truly kernel-agnostic GNU system? 'Cuz if it's the latter, well...
*cue John Cleese quote*
Offline
I myself really like the HURD hybrid-kernel idea but am trying to be patient and wait for the true "stable release", or whatever they want to call it when it's ready next year or so. In the meantime I was also looking at other projects with similar ideas such as Gentoo's BSD ports (Gentoo/{Free,Net,Open}BSD) where you use the Gentoo administrative tools (portage, etc) and scripts but built ontop of a BSD kernel and toolset. A good possibility for what I may personally enjoy is USE="static static-libs" and install `musl' and have the `musl-wrapper' symlinked for builds. More of an experiment, I guess, but in the direction I am looking for since I really really enjoy `emerge' and friends. Again, I still want to start from scratch but the ports look interesting enough to play around with. I am so picky I think I may box myself into a difficult to maintain setup... but who knows.
Still waiting to see how your project turns out since you seem to be further along the path than myself.
Edit: Anyway, I apologize if I seem like I'm brainstorming or simply rambling. I just get a bit excited during these in-depth experiments and want to discuss ideas further to see what others think of the subject.
Last edited by milomouse (2013-08-20 22:59:08)
Offline
Just found this as the top GNU/Linux Discussion topic; did it ever go anywhere?
Offline
This thread is three days old, friend. Unless I misunderstand you; it's pretty late here, I'm tired.
Offline
Oh, silly me. I'm new (and it's late) and I read people's registration dates as the post date so I thought it was from 2009... *facepalms*
Offline
I myself really like the HURD hybrid-kernel idea but am trying to be patient and wait for the true "stable release", or whatever they want to call it when it's ready next year or so
Be prepared for the eventuality of "next year or so" being pushed off to "the year after that or so". Wash, rinse, and repeat. They've been saying it for a bit now.
That's not to say that the Hurd doesn't hold promise, or that microkernels in general don't hold promise. But the Hurd has managed to not deliver a stable system in 23 years of development. Now I haven't tried it so I can't say with any authority what the status of the project is, but from my understanding the Hurd is still mostly missing support for character devices (including sound cards) among other deficiencies. That's enough of a show stopper for me.
In comparison minix3 was announced in 2005 and is a mostly complete OS right now. The bulk of the work being done now is in porting the NetBSD userland and getting more of PKGSRC to compile and run successfully. It's absolutely not GNU, but it seems to be making a lot more progress than the Hurd from what I can see.
Personally I've reached that state of mind where I'm not very interested in distro hopping, trying to build my own system, or looking for the next big thing. I spent a lot of time on Linux From Scratch and automating the build with Makefiles, which was pretty cool in a way, but to get a truly useful system the amount of work made it feel unmaintainable. I can't say definitively that Arch will be my final OS install as I still love FreeBSD, but it's as good an OS as I've ever run and I'm not interested in changing any time soon. I'd rather stick with it and use my computer to get actual work done. It's always up to date requiring minimal intervention and damn stable for being right on the cutting edge.
Offline
The L4 microkernel family seems to be an interesting development. I love the idea of microkernels and I've seen a nice system execution in the form of Microware's OS9 (unfortunately not free and these days more a historical curiosity than anything). Hurd doesn't inspire huge confidence - it was announced in 1993, the first time I ever used Linux (Slackware, installed from over 9000 floppy disks, downloaded via 57k state of the art dial-up.
Offline
Pages: 1