Slashdot Mirror
Linux From Scratch 6.0 Released
Bubblehead writes "The Linux From Scratch community is pleased to announce the release of
LFS 6.0. This major revision of the book includes a number of major
package upgrades, including GCC 3.4.x, Linux kernel 2.6.8.1, and the
Udev software package, allowing for dynamic creation of device nodes.
The text has also been vastly re-written for improved readability."
Why use this? Isnt package management and things like Debian's Dpkg, rpm and stuff are for?
That and I know nobody who uses this.
I wanna read Linux from Sniff tho.
If you like what I've said here, and want to read more, go to http://www.krillrblog.com
As a distro, LFS probably isn't what most people want. If you want to compile from source, then gentoo is probably the way to go.
On the other hand, I dont think there is a better way than LFS for learning Linux and what all the different packages are for. It is a good way for linux newbs (but probably not computer n00bs) to learn about their new OS. Once that's done, head over to an easier-to-maintain distro such as debian.
just what we need, more linux distributions :-).
"The text has also been vastly re-written for improved readability"
Now if only they'd do that to the kernel source...
If you like what I've said here, and want to read more, go to http://www.krillrblog.com
If you build this distro yourself, how can it be updated?
I started with a Slackware distro using the 0.96 kernel, largely archived onto floppies and had to feel my way through making it boot, so I consider myself as having learnt not quite from scratch. But today's folks have it easy, thanks to people who already understand what's under the bonnet.
I bet most people who LFS will gain a better appreciation for configure-and-compile-yourself. Three cheers!
Smaller user base? Less support? less market share? If you want linux for wusses, install Fedora, if you want to do it ALL from scratch, use this, and if you want something in between, gentoo works well.
What's the problem?
Linux sucks anyway. It's all about the BSDs.
If there's a joke there, I missed it.
A long time ago, when I was more confused about Linux and kernels and other big words that involve penguins and popcorn, I set up a completely minimal Mandrake base system and then built everything else from the LFS instructions. I think that intermediate level gave me a better grasp of what I was doing.
I don't know if I'd recommend everyone try it, but it sure was an experience I found valuable when later having to mess with even binary package-based distros like Debian.
Sig!
I'm on my second installation of LFS. It's not for those in a hurry, that's for sure- It takes me about a month to get a VERY basic system. (Think X and Firefox, and not much else!). I inevitably make some bone-headed mistake, and I don't back up while installing, so it's my fault.
:) (And as close as I'll ever get)
Also, building something like gnome from source really teaches you the meaning of 'dependency hell.' Seems like every single package requires exactly version "1.1211-1243pm" of every OTHER package, and even some obscure RedHat patches too, in some cases. *Shudder*gDesklets*Shudder*.
However, I have a VERY good idea of what's on my system and what libraries are being used where. There's NO 'fat' or extra stuff that I didn't put there. I used to shy away from compiling from source- which makes sense on a package-based system. Now there's NO fear. There's even a few interesting package management schemes for compiling from source.
Booting into my brand new, hand-built, bare-bones system- it's almost as good as if I'd written the software myself
It basically becomes your perogative as to what gets updated and when.
If you're wondering what LFS's primary use is though...or what it seems to get used for a lot, is the creation of new distributions. People will build an LFS system, and from that you end up with Yoper or GoboLinux, to name but two.
I know a lot of people seem to have difficulty understanding the value of new distributions, but there are many reasons why they are valuable. The first is that for people who are sufficiently technically inclined and proactive, if none of the existing distributions fits their needs for a given purpose, (and yes, it still can and does happen) LFS gives them the ability to put together something exactly the way they want it...with everything they do want, and nothing they don't.
Another benefit of this system is that it encourages people to be self-sufficient, rather than relying on corporations to provide what they need...corporations who generally care far more about their own interests than those of the user anyway.
Yet another plus is that it stimulates and encourages technological progress. I've covered this topic before, but anyone who has read Darwin will know that in order for anything to advance according to the evolutionary model, there needs to be a lot of different instances of a given thing...the process needs to experiment with a lot of different mutations before it is decided which mutations are permanently integrated into new generations of the organism. The more different distributions and forms of Linux exist, the more this process in encouraged.
I think the reason why people dislike the idea of new distributions is because they look at things from a Microsoft-like perspective of usability, which unfortunately involves a couple of extremely negative assumptions.
1) That the end user is a drooling imbecile, who needs to have things made easy to the point of them being rote. Intellectual participation in computer use is seen as more anathema than anything else.
2) Because of the deep level of retardation that is assumed in the end user, it is therefore also assumed that the level of usability exists in inverse proportion to the level of diversity. That is, in order to keep things usable it is necessary to minimise the number of different possible solutions to a given problem, or software programs, as much as possible in order to avoid users becoming overwhelmed.
The problem is that if these two points are adhered to and followed, a number of other very bad things happen. One is that technological advancement grinds to a screeching halt, as we have seen in the current state of Microsoft's software. Because innovation is very difficult when these two points are adhered to, we then get security problems of the kind that we have also seen.
The other bad thing that this causes is that it promotes the idea that intellectual laziness is not only acceptable, but that it's actually a good thing...when the opposite is in fact true.
People need to realise that having new distributions isn't going to by definition hurt anyone, and that it is actually very good for Linux as a whole. If you only want to use one distro yourself without deviation, that's fine. But IMHO it is wrong to try and impose your own desire for uniformity, lack of diversity, and stagnation on the rest of the world.
Yet there is a group of people who do this.
I knew one couple who had actually had their own house build according to their design but the husband who was my collegue couldn't understand why I didn't like the read made windows desktop.
LFS is the most extreme way to get a linux system unless your firstname rhymes with a popular OS. What does it do? Well nothing except really show you what makes up a system. It won't teach you anything about programming but it will teach you a lot about just how complicated a modern computer system really is. The amount of code needed to create even the most basic system is insane. Start adding stuff like a grahpical desktop and you might start to have a better understanding of how all the software packages work together to make your desktop.
It is like stripping a car engine. Doing it doesn't make you a better driver but it can be a rewarding learning experience nonetheless.
Some of us are not satisfied to work with blackboxes. Just like those people who want THEIR house to be THEIR house LFS allows you total control over your system. Of course most users use a "regular" distro for their actual work but just maybe they have come to understand their systems a little better.
But to answer your question directly. If you got to ask why then it is not meant for you. It is like asking why people climb mountaints. Because they are there.
MMO Quests are like orgasms:
You may solo them, I prefer them in a group.
The youngsters have it too easy. When Redhat hadn't invent rpm, you had to get the tarball or check the source out of the cvs and try to get it compile, for every sodding package. That's educational. Gentoo isn't. FreeBSD portage system isn't. They are convenient way of building from source but they are not that different than rpm-hell in many ways.
To put it bluntly, when I was young, we had to walk 10 miles in snow everyday, both ways uphill. :)
there's Gentoo!!!
Donald 'Duck' Dunn: We had a band powerful enough to turn goat piss into gasoline.
I do totally agree. LFS is for the ones who want to learn all about the system. Who wants to educate oneself about GNU/Linux, should read it.
I built a LFS about 6 months ago from the 5.x series. It was a great learning experience even though most of the time, I was just following instructions. I immediately gained much more respect for the packaged Linux distributions, not because LFS was bad, but because of the detail and complexity involved in putting to gather a working system.
My first suprise was the level of patching and configuration.I knew that some people described linux as a "hodgepodge" of different components and not a complete system. I always thought that was slightly trollish, but, in a way, I have came to appreciate what they mean. Most especially, I was suprised at how many of those components do not work togather unless they are patched or configured to force them to live togather. You can't just download the vanilla source of a package, compile it with defaults, and expect it to integrate into Linux. Lots of the packages had to be patched to make them work in Linux. Also, some of the packages have to have slightly esoteric configuration options set. If I hadn't been reading the instructions, I would have never known what patch to install or what configuration option to set. This gave me a lot of repect for the people who make distributions because they had to work that out for themselves and know all that stuff in detail.
My second suprise (and consequent appreciation for package distributions) has to do with the kernel patching. I was using an old Athalon 750. With a vanilla 2.4 kernel, it throws strange "spurious" interrupts on IRQ 7. But a Redhat 2.4 kernel fixes it. That tells me that the Redhat people took some extra effort to fix those annoying issues.
So in addition to being a great learning experience, LFS made me appreciate the effort that Redat, Debian, et. al. put into their work.
Here.
It's a simple idea, but works well. The FAQ there will explain it all.
You are the first zealot to speak up about your distro in a thread that has NOTHING TO DO WITH IT!!
Huge surprise, it's Debian you mention.
FreeBSD's portage is not much different compared to Gentoo
I shouldn't have to explain what is inherantly wrong with this statement.
For every annoying gentoo user, are three even more annoying anti-gentoo crybabies. Take Yosh from #Gimp for example.
The followup book Beyond Linux from Scratch goes way beyond compiling a basic system. It has all the instructions and patches needed to get X11, KDE, Gnome, Office Suites, etc compiled from scratch. The 2 books have matching version numbers for compatibility.
I've built a few of these. They work better than any other Linux system I've seen. Mostly, this is because by building them you learn how to fix something when it doesn't work.
There are times you want to patch, configure, and compile from source, and there are times when you just need a working system ASAP. That doesn't make either one wrong.
Yet another plus is that it stimulates and encourages technological progress. I've covered this topic before, but anyone who has read Darwin will know that in order for anything to advance according to the evolutionary model, there needs to be a lot of different instances of a given thing...the process needs to experiment with a lot of different mutations before it is decided which mutations are permanently integrated into new generations of the organism.
The difference is that every change made to a distribution or piece of software is intended to better the software. With Darwinism, each change not only occurs very rarely, but you also need the change to happen in the right conditions (good competition) and the change actually has to help the situation significantly. And the change has to be significant that it doesn't get watered down in the gene pool of the rest of the group.
Darwinism has been dumped by scientests in favor of punctuated equilibrium--which suggests (amongst other things) that mutation only drives change when random external events like meteors, sudden tribal encounters (in the case of homonids developing cereberal cortexes), etc. enter the picture.
I don't know how a system that progresses itself by accident can be compared to a system that progresses by deliberation.
The progress of distributions can be better compared to the advancement of humanity in societies.
Slashdot: Where people pretend to be twice as smart as they really are by behaving like children.
A nice side effect of going through the LFS book is that you get to see just how poor the GNU autotools really are. Sure they're better than other schemes, but it takes damned little imagination to figure out how they could be better.
/usr/bin is retarded. DragonFlyBSD is doing something nice to solve this problem, and their solution will work even when these program designers get their shit together.
.mozconfig file, and it's possible to get a working Firefox with any number of a combination of options, but I doubt that a canonical list exists anywhere. How the fuck are people expected to text a program when they don't know what different possible combinations of a program exist? (For the record, I'm using Firefox-1.0 now, and have successfully built it from source, using a cobbled together .mozconfig file that gave terrific results. I'd hate to browse the Internet without it, but I also hate building it expressly because of a lack of detailed .mozconfig options).
The same goes for the FHS layout, and you'll realize this the first time you need to install a program that requires autoconf-2.54, but you installed 2.59. Different versions don't always play well together, and configure scripts can be dicks about what versions they'll allow. Mixing programs together into a blob in
Finally, the greatest side effect to going through LFS is that you get to see just how pathetic the documentation for some source code packages really is. You could not write the LFS book as a list of pointers to other peoples' documentation, because it doesn't exist (not that the LFS people would do this--more on that later). Some people will go through the trouble of creating an astounding, outstanding program that's destined for ubiquity, but they won't bother to include a list of dependencies. You don't find out you need X and Y until autoconf bitches about not finding them. Other people will provide decent documentation for running a program, but won't provide even passable installation instructions. Mozilla Firefox is a terrific example of this. There are dozens of options that go in a
LFS is an astounding project, because it's the best documentation that exists for some of the software it includes. It's quite complete in listing what compilation options do, and what the ramifications of those options are. I used LFS-4.0, and would probably have stuck to LFS, except for my inexperience as a UNIX/Linux admin. I had holes in how I was supposed to manage parts of the LFS system I had, especially when it came to upgrading. I jumped to BSD, just in case anybody was wondering (which I doubt).
Anyway, I applaud the LFS crew. There's no doubt that what they're doing truly amounts to an increase in understanding of Linux in the world.