Yet Another "Can I make my own Distro?" Thread

Yes, there comes a time, once a week where someone needs to ask this question. This week I take the torch and run with it.

So yeah, my question basically stands in the thread title. I'm looking to make a distro - not just for myself, but customize it for a web community I go to. I was debating either doing this or modifying Ubuntu. I feel like if I made a LFS system that I could distribute, it'd be more satisfying -- it would be more "ours", where as redoing Ubuntu would be easier now and in the long run.

But still, I'd like to know: If I were to make a running LFS system, is there a way to get it from point A to point B? I.E. CD to Hard Disk? Would it be possible to write some sort of installer?

The short answer is "yes".
The question is wayy too open ended for a long answer.

<sigh> oh all right: here's some starting points...
http://www-128.ibm.com/developerworks/library/os-lfs/
http://www.cyberpunkcafe.com/news.php?extend.83

Take a look at all the distros listed at distrowatch and ask yourself if you can provide something fundamentally different.

I'm sitting here trying to get a laptop running with Linux (wireless, power management, etc.) All the distros in my stable have pretty much common features as respects laptops---the LAST thing I need right now is a new distro.

In my business we constantly ask engineers what problem they are trying to solve---its amazing how many can't answer that.

Originally Posted by pixellany In my business we constantly ask engineers what problem they are trying to solve---its amazing how many can't answer that.

I totally agree with you this time but I hate the idea behind that statement.

In my long engineering experience I've worked with a lot of people who are too aware of the stereotype of engineers, including the "solution looking for a problem" behavior. Those people would say exactly what you said, but the real meaning behind their "its amazing how many can't answer that" was "I already knew the engineer was wrong before I asked, so I didn't bother listening to the answer".

My apologies (to both of you) for stating snap judgments that might have been better left silent. But my snap judgment on testforechozero is a notch worse that "solution looking for a problem". He doesn't quite have the solution yet and he might not even be looking for the problem. I think making snap judgments (not necessarily airing them) is important to being a good engineer. Discarding them as soon as better information comes in is even more important.

Testforechozero, if you care that you're giving that "solution in search of a problem" impression (no reason you need to care). You might explain the problem you intend to solve. What will your distro do that's better than the existing ones? Doing so might even improve the tone and quality of answers you get.

Well you should start off with thinking about the thing from top to bottom, like usually. If you don't find a distribution that "suits you", mostly by it's outlook and out-of-box-installed apps, the easiest way is probably to remaster an existing, good-feeling distribution. If you want a fundamentally different distribution for your purposes, it might be worth it to make it "from scratch". You said you know it's easier, especially in the long run, to remaster Ubuntu than always make a Linux From Scratch or equivalent, but you should know the magnitude of it: creating a new distribution isn't just about downloading new packages, or rather source code, then configuring/compiling them and packaging the product onto a cd along with an installer. There's a whole lot to do - decide what is important not only for you, but for all the others with varying hardware, not just yours; check new versions existence, their new dependencies, hunt patches for them when they don't work or need fixing, ensuring that the previous distribution you made is compatible with this new one so that the users moving on don't have to tear their hair off with the new one. Slackware's release cycle tells about this: new version is released (or so it's said) when it's ready, not every six months (Ubuntu does this the other way around, and while it's ensured there's new software every six months, it's another matter how stable it is, or if there is anything really changed except for version numbers) - and mostly the Slackware releases are considered stable. Not perfect of course, but stable.

You should, if you still haven't, read and make your way trough the LFS book first. And I mean read it, not just issue the commands or do it semi-automatically; understand it. Then when you get it running and have a grasp of it, maybe do it a second time, but this time making it not exactly by the book - for example downloading newer sources, looking at what they depend on (and getting the dependencies as well), making them compile and install - to get a grasp of what it is when you have no book to follow, but hungry mouths waiting for new things to chew. Then try to remaster Ubuntu to add your own wallpaper and a few new preinstalled apps, and ask yourself which one you prefer - did either method satisfy your needs concerning the new distribution?

I consider it beneficial to try, because it's the best way to learn. Only after trying you would really know which one, if either one, is your way - it is true that there are loads of distributions already and that it isn't of much use to make your "own distribution", but on the other hand, if you never try you never find out what it's like.

Since source code, distributions, live-cds, installers and such exist, it is possible to
- remaster an existing distribution
- create a new distribution from scratch
- make a live-cd of your new operating system
- make an install media of your new operating system
- make it better than the others (for your needs) and have it become popular
- fail to make it good enough for anyone except for yourself and bury it in silence
- learn a lot about the process and get new ideas

So: prepare some space on your harddisk and start finding out how it works.