I did LFS once, a long long time ago (days of RedHat 7.3, I'd say, 2002), and it was quite educational! It's a good way to learn the fundamentals of what comprises an actual Linux system
Took me half a second to realize I would probably start with "LFS :: Linux From Scratch is the main book, the base from which all other projects are derived."
My journey was: Start off by clicking LFS among the links, and then "Read online" in the sidebar, and then being lost. I see a header, "Current Stable" but I can't see the link to the actual book. I only see the errata, the security advisory and a link labelled... stable LFS systemd? Maybe it's that one? Yes, yes it is.
For a long time (probably 2002-2010) I really wanted to do LFS. But I've been on Gentoo since 2004 which, over the last ~22 years, has probably exposed me to almost everything LFS would have, plus a lot more it wouldn't have. I still wish I'd did it way back when. I probably would stumbled through fewer issues had I.
I had the same experience. I think the only thing that LFS would add to that experience would be the package-manager-less experience (building without portage's help)
LFS takes the wrong approach in my opinion. I did it once manually over a couple of days. I suppose I hoped enduring the pain of waiting for the packages to compile would somehow make me smarter. It didn’t. I didn’t learn much. It was like following a baking recipe. I know I need to add eggs and sugar in the ratio it says but I’d be damned if I could tell you why that specific ratio works.
It’s just too much to take in at once. You start by building a cross compilation toolchain but… why? And you add the bootloader dead last, even though it’s the first thing you see when you boot a Linux machine?
I reckon a better approach would be to build it step by step, only adding new things once you’ve discovered a need for them. First you get the bootloader working, then you get it to actually load the initramfs and kernel, etc… you could even create your own extremely simple initramfs first before adding dracut, etc…
This gets you to a system you can boot and play with asap. And then you can slowly start to discover why you actually need all these things through exploration and experimentation.
I've learned the most from LFS long time ago, when I had no internet. I had a mismatching set of Slackware and redhat source CDs, and LFS itself didn't match the versions on CDs too. Since then I can build anything and anywhere, even if it's not cooperating :)
You know, you could also just read about each step to know more. I mean if you feel you're following a baking recipe then you can just... read about it more. And then you have LFS to have a simple way of experimenting whether you want to change anything or not. Also technically you don't need to have a bootloader installed on your main hard drive if you already have it on other disk.
I mean eh... last time I've checked this was Hacker news. Yesterday there was an Ask HN topic how to enable dark mode on HN. Is this the new hacker culture?
I learned a lot doing Gentoo stage 1. I was also following the guide like a recipe, but at the time I hadn't been exposed to the underlying machinery that is often automated, abstracted away, or done for me already. It was also a good test on how well I could follow instructions and later on apply my own customisations. Like lego.
I second this a thousand million times. But with AI, i can't help but admit i find myself indulging in building things backwards, now that i can try a couple different approaches simultaneously for it.
As pointed out already, LFS is a cookbook, and you learn most by having an idea about how you want to tune your Linux system.
So at points you want to dive deeper, you do — and go off the beaten path.
It requires a desire not just to understand how bits and pieces fit together, but how you can make them work even better for yourself and what is and is not customizable?
Arguably, even if building it today is easier — back when I did it in 2000, you sometimes struggled to find where the canonical source code of the package is, not to mention compilation errors, linking errors, etc. Only then did you get to configuring the init RC system, choosing between them (no systemd yet), etc.
But a requirement is — IMO — that you are strongly opinionated about how a system should serve you, and look for opportunities to tune it.
This should be said more often, and in many other contexts as well.
Most educational resources are what- and how-based, only the best ones are primarily why-based, what Wirth once wrote in the preface to some Pascal book (I paraphrase) 'discussions are motivated, so that the learner gets coverage and understanding, instead of memorization'.
I used CLFS (Cross-Linux From Scratch) once to bootstrap Linux on my SGI Indy, and I learned quite a lot from this. That was back in 2008 or so, so way before AI and extensive online guides for almost everything. It taught me how cross compilers work, what dependencies exist in the GNU toolchain, how the SGI boots, and much more.
Too bad Cross-LFS is no longer a thing. Then again, with all these microcontrollers and ARM and RISC-V systems today, cross-compiling is no longer as magic (and tricky) as it was back then...
LFS helped me a lot. It taught me how to compile packages manually, and how GNU/Linux components fit together.
Armed with that knowledge, I could do things others could not do:
- Make applications that bundle/automatically download their dependencies, compile them with custom flags (static linking)
- Diagnose build errors buried in deep dependencies when using vcpkg
- Compile programs for esotheric systems: e.g: Intel MacOS X 10.5, an obsolete system noone else on earth is using except the company I worked at
- Create application targetting obsolete systems but using the latest version of its dependencies
- Compile recent programs on esotheric systems, by bootstraping a newer gcc on the newer system, patching as necessary
- Create my own stripped down distro, tuned to my specifications
- Use that stripped distro as a docker container
- Run recent linux on obsolete hardware no longer supported by mainstream distros
- Reformat a cloud dedicated server without physical access by first bootstraping a minimal distro on a ramdisk, unmounting the main drive, partitioning that drive, and installing the new distro on that drive. Reboot and pray.
- Build a cross-compiler for targetting a freestanding environment
- Build latest KDE/XFCE for debian stable. No more "but debian ships with outdated software!!11"
- Build applications on a distro they were not designed for (Zimbra OSE on debian)
None of that requires LFS; I compiled KDE via scripts many years ago. Why would LFS be needed for that? That makes no sense. Debian being slow is their problem, other distributions aren't snail-like like that. The default compiler used in LFS is also hardly a real cross-compiler, it is more a bare-minimum compiler for C/C++. This annoys me because e. g. I need fortran for R, so a bare-minimum compiler isn't as useful as a more feature-complete GCC. And GCC also causes some issues, e. g. like my problem with options.cc.
I think you missed the main argument though - LFS isn't quite as good as it could be. That is a correct statement that was made.
Learning how to compile packages manually it was a common skill long before Linux came to be, and how GNU/Linux fits together could also easily be learnt with Slackware or Yggdrasil.
>Learning how to compile packages manually it was a common skill long before Linux came to be
Right. I used to compile many open source packages from source on many flavors of Unix back in the day, for years. Perl, Python, MySQL, many others. Miscellaneous utilities. Esoteric languages. Even Oracle, once, on a Unix server, right at a large manufacturing company site. Even building Unix kernels from source was common on some Unix versions. And not just for a regular install. Many times, (re)building Unix kernels from source, was the only way to get some software packages or hardware devices or peripherals, like printers or terminal controllers or even hard disk controllers, to work on the system. Typically you would edit some config files or settings files manually, change values of params, or in some cases use a TUI to do that, and then rebuild via a makefile. You had to save the old kernel in case the new one didn't work, and if so, lather, rinse, repeat, till everything did work.
And that practice was nothing special. A lot if people did it routinely. Most, in fact, I'd say. Talking about devs and sysadmins here, not end users, although some of the latter probably did it too.
Many apps, tools and utilities came as source code, often in a .tar or .tar.gz file. You would untar or gunzip it into a new directory, and do your stuff, running makefiles, other shell commands, setting env. vars., etc.
As someone who started using Linux in 1996 (Slackware, of course), I learnt the hard way about compiling packages. I remember trying to compile the kernel, and doing "make config" and following hundreds of options, and not knowing that I then needed to do "make" or "make zimage" or "make install" after any of the config stuff. Obvious, dumb, rookie errors, but still understandable for a teenager coming from Windows. Doing LFS really taught me a lot about dependencies and complexity, as well as understanding of the actual boot process. All of this helped me greatly in later years.
> And you add the bootloader dead last, even though it’s the first thing you see when you boot a Linux machine?
That actually makes perfect sense. If you did the bootloader first, how are you going to build it? How do you know what to boot from which filesystem on what device? By the time you've got your root filesystem complete, it's easy to tell.
If you do things step by step incrementally, you keep having to go further back in the process. Starting over repeatedly gets tedious, boring and frustrating. And it'll take ages to finish the complete OS.
Except that if you fail to install the bootloader - which happened to me many times - then you do not have a working system. So really the prior argument makes sense.
IMO the guy is right - LFS could be way better. But it has this current structure and does not deviate from it. That's the bad part.
> If you do things step by step incrementally, you keep having to go further back in the process. Starting over repeatedly gets tedious, boring and frustrating. And it'll take ages to finish the complete OS.
Pray tell how that is necessary for the bootloader? The chain used in LFS isn't the only one that makes sense or is required. IMO it would indeed be better to have a system you boot into as quickly as possible, on real hardware, then continue from there.
If you're going to run the OS, you have to make sure it compiles and runs the bootloader and its installer. The toolchain could be different enough for that to fail. And you do boot into a reasonably small OS and continue from there. That's what BLFS is all about.
Agreed, it's near the end of the Gentoo installation as well. If you forget to install GRUB, just chroot back into it and continue.
If I installed the bootloader first, I'd have to remember to run dracut after the system was installed. Or I install it last, create the init at the same time and make a habit of it.
Sounds like a speedrun of the tutorial. How did you approach packaging then and what system did you use to compile everything?
I did the tutorial in the late 2000s I think and it took me 3-4 months if I remember correctly. At the time I wasn't completely new to Linux but setting up things like the boot loader was quite tricky for me.
For me the main motivation was getting a deeper understanding, being able to fine tune and customize my system. Of course I didn't go blindly through the tutorial but only did steps that I understood. (In case something goes wrong) Also I started to use (x)stow for everything.
It is possible to do it over a few days really, even with not too much knowledge.
Requiring four months seems as if it was a side project, which is of course totally fine, but for any main task it really should not take more than a few days, even on a semi-slow computer (well, depends on how slow).
To some extent I agree with you. I'd like to decouple some parts,
e. g. the tutorial part is great but it should ideally be distinct
from "just go at it" 1:1. I also fail to compile various things right
now, which is frustrating; two or three years ago all went well. The
whole LFS build up is too brittle. In part this is due to external
complexity, but LFS needs to re-think what it does. It's also sad that
they are now systemd-only, which further adds intrinsic uselessness. I
don't need anything systemd offers; all my boot up logic is stored in
yaml files and autogenerated to target formats via ruby as-is.
well, if you happen to be in a 'rm -Rf /' situation (which with agentic AI seems to be more often the case) to know how to operate in a chroot environment could be helpful
> I did it once manually over a couple of days. I suppose I hoped enduring the pain of waiting for the packages to compile would somehow make me smarter. It didn’t. I didn’t learn much.
Exactly my experience. Back in the day when all I knew was Windows and a couple of BSDs I wanted to get into Linux, because it seemed like it was going to become very relevant in the next decades. After years of slacking I thought LFS would probably be the ideal choice. Turned out it wasn't, at least for me. I ended up "learning Linux" with distros like Gentoo, Arch and later even Ubuntu, and by just replacing my production systems with Linux, which were still running on OpenBSD mostly. I have revisited LFS a couple of years ago and honestly didn't take much from it. I get the appeal, but it's not for everyone.
Exactly my experience. I tried LFS several times, but I didn't learn much. I learned more with Arch Linux, but I truly appreciate the work that was put in LFS, marvelous for the right type of learner.
The step by step is the right approach... but you'll discover that many 'components' are _INSANE_ and that includes their SDK.
Bare LFS is not enough (I run my own): you need a kind of userland (above glibc) multi-version system which some kind of "atomic-ish" switching (always have a stable SSH running in case something goes wrong, better than unplugging the system disk and fix it on another computer).
For linux, same thing, with a kind of flip-flop-ing for updates/fix/etc.
The insight that gives you will probably scare you: the current "open source" stack is an abomination (the worst is the SDKs I think).
That's why we need _LEAN_ open source, and that includes the SDKs.
If you think that's an abomination, try corporate billed-by-the-hour enterprise software where nothing ever changes unless it's approved by several layers of management.
I promised myself not to work for any of those anymore :)
This whole project wastes time that could be spent learning operating system design and it spends it telling users to download files and run ./configure; make; make install 100 times.
I do LFS every year, just to keep the chops sharp. It has been an immensely invaluable means of keeping informed with the way things are rolling in the Linux world.
I kept an LFS system going for a few months (on a wretchedly awful by modern standards celeron “net top”) until eventually making a mess I couldn’t figure out how to untangle during a system upgrade. this gave me the sheer force of will necessary to later do things like own the multiplatform build system for an ocaml based distributed system (mina)
LFS is great. Unfortunately nowadays I often fail to compile
it; most recently with GCC complaining and failing at options.cc.
But I also get more issues here.
I kind of need a LFS variant that is more flexible. For instance,
the part up to having a working standalone compiler, should ideally
become more robust and/or decoupled from the actual LFS build process.
Once I have a sane toolchain, the rest is trivial, ruby scripts solve
that part for me. But the build up towards that is not so reliable.
Things also fluctuate so much in the build process; I still need perl
for instance, and python wants to add more and more complexity into the
build system as well, even though meson/ninja is a nice combo. For some
reason, the build situation on Linux is not very elegant to word this
nicely. Then we also "now you need LLVM and glslang and spirv" and so
forth, and so on. The complexity always increases. To some extent I
can understand why slackware crumbled. How can a single person, even
with scripts, keep that all working?
LFS is worth doing if you are interested in learning how Linux distro is put together. There is a fair bit of reading involved. I did the entire thing in a Debian VM.
I m very happy I started years ago with Gentoo which is also from scratch but with a package manager ones your system is up ( couple of days). You learn so much just starting from zero and understand choices you can make on that journey. In the end I stopped using it though, way later when the Gentoo hype was over though. Getting stuck into dependency loops wasn't productive anymore.
I remember using LFS’ predecessors to build 1.44MB “rescue” floppy disks with Linux 2.0 in the 1990s. Tweak the config so it could fit on there along with a small userland. I don’t even remember if busybox was around back then.
It used to be a teenager’s rite of passage to build your kernel config, figure out LILO (remember that, pre-grub!), ipchains.. talk about feeling old!
77 comments
[ 3.1 ms ] story [ 23.5 ms ] thread> Bruce Dubbs - 2026/09/01
> The Linux From Scratch community announces the release of LFS Version 13.1.
It’s just too much to take in at once. You start by building a cross compilation toolchain but… why? And you add the bootloader dead last, even though it’s the first thing you see when you boot a Linux machine?
I reckon a better approach would be to build it step by step, only adding new things once you’ve discovered a need for them. First you get the bootloader working, then you get it to actually load the initramfs and kernel, etc… you could even create your own extremely simple initramfs first before adding dracut, etc…
This gets you to a system you can boot and play with asap. And then you can slowly start to discover why you actually need all these things through exploration and experimentation.
I mean eh... last time I've checked this was Hacker news. Yesterday there was an Ask HN topic how to enable dark mode on HN. Is this the new hacker culture?
Principles can be studied from “Operating Systems: Three Easy Pieces“ book - https://pages.cs.wisc.edu/~remzi/OSTEP/
Rob Love wrote the book “Linux Kernel Development”. https://rlove.org/
And then continue with reading LFS.
So at points you want to dive deeper, you do — and go off the beaten path.
It requires a desire not just to understand how bits and pieces fit together, but how you can make them work even better for yourself and what is and is not customizable?
Arguably, even if building it today is easier — back when I did it in 2000, you sometimes struggled to find where the canonical source code of the package is, not to mention compilation errors, linking errors, etc. Only then did you get to configuring the init RC system, choosing between them (no systemd yet), etc.
But a requirement is — IMO — that you are strongly opinionated about how a system should serve you, and look for opportunities to tune it.
Shameless plug: I started a series of blog posts exactly because this. https://serversfor.dev/linux-inside-out/
Unfortunately due to some cervical spine issues and a surgery I wasn't able to finish it, but I am planning to pick it up and continue.
Most educational resources are what- and how-based, only the best ones are primarily why-based, what Wirth once wrote in the preface to some Pascal book (I paraphrase) 'discussions are motivated, so that the learner gets coverage and understanding, instead of memorization'.
Too bad Cross-LFS is no longer a thing. Then again, with all these microcontrollers and ARM and RISC-V systems today, cross-compiling is no longer as magic (and tricky) as it was back then...
LFS helped me a lot. It taught me how to compile packages manually, and how GNU/Linux components fit together.
Armed with that knowledge, I could do things others could not do:
- Make applications that bundle/automatically download their dependencies, compile them with custom flags (static linking)
- Diagnose build errors buried in deep dependencies when using vcpkg
- Compile programs for esotheric systems: e.g: Intel MacOS X 10.5, an obsolete system noone else on earth is using except the company I worked at
- Create application targetting obsolete systems but using the latest version of its dependencies
- Compile recent programs on esotheric systems, by bootstraping a newer gcc on the newer system, patching as necessary
- Create my own stripped down distro, tuned to my specifications
- Use that stripped distro as a docker container
- Run recent linux on obsolete hardware no longer supported by mainstream distros
- Reformat a cloud dedicated server without physical access by first bootstraping a minimal distro on a ramdisk, unmounting the main drive, partitioning that drive, and installing the new distro on that drive. Reboot and pray.
- Build a cross-compiler for targetting a freestanding environment
- Build latest KDE/XFCE for debian stable. No more "but debian ships with outdated software!!11"
- Build applications on a distro they were not designed for (Zimbra OSE on debian)
I think you missed the main argument though - LFS isn't quite as good as it could be. That is a correct statement that was made.
LFS is about the muscle memory; the documentation; the cohesiveness.
I always recommend it to people and consider it foundational to my understanding of How Stuff Works.
Whether that amounts to a hill of beans in the AI world remains to be seen.
Fundamental things like how processes are scheduled are not explained by LFS.
Right. I used to compile many open source packages from source on many flavors of Unix back in the day, for years. Perl, Python, MySQL, many others. Miscellaneous utilities. Esoteric languages. Even Oracle, once, on a Unix server, right at a large manufacturing company site. Even building Unix kernels from source was common on some Unix versions. And not just for a regular install. Many times, (re)building Unix kernels from source, was the only way to get some software packages or hardware devices or peripherals, like printers or terminal controllers or even hard disk controllers, to work on the system. Typically you would edit some config files or settings files manually, change values of params, or in some cases use a TUI to do that, and then rebuild via a makefile. You had to save the old kernel in case the new one didn't work, and if so, lather, rinse, repeat, till everything did work.
And that practice was nothing special. A lot if people did it routinely. Most, in fact, I'd say. Talking about devs and sysadmins here, not end users, although some of the latter probably did it too.
Many apps, tools and utilities came as source code, often in a .tar or .tar.gz file. You would untar or gunzip it into a new directory, and do your stuff, running makefiles, other shell commands, setting env. vars., etc.
I certainly do not miss compiling X11!
That actually makes perfect sense. If you did the bootloader first, how are you going to build it? How do you know what to boot from which filesystem on what device? By the time you've got your root filesystem complete, it's easy to tell.
If you do things step by step incrementally, you keep having to go further back in the process. Starting over repeatedly gets tedious, boring and frustrating. And it'll take ages to finish the complete OS.
IMO the guy is right - LFS could be way better. But it has this current structure and does not deviate from it. That's the bad part.
> If you do things step by step incrementally, you keep having to go further back in the process. Starting over repeatedly gets tedious, boring and frustrating. And it'll take ages to finish the complete OS.
Pray tell how that is necessary for the bootloader? The chain used in LFS isn't the only one that makes sense or is required. IMO it would indeed be better to have a system you boot into as quickly as possible, on real hardware, then continue from there.
And you dont need to boot anything, just a busy loop after a log "Hello World" is enough, if you feel fancy, maybe make a simple shell ?
If I installed the bootloader first, I'd have to remember to run dracut after the system was installed. Or I install it last, create the init at the same time and make a habit of it.
Sounds like a speedrun of the tutorial. How did you approach packaging then and what system did you use to compile everything?
I did the tutorial in the late 2000s I think and it took me 3-4 months if I remember correctly. At the time I wasn't completely new to Linux but setting up things like the boot loader was quite tricky for me.
For me the main motivation was getting a deeper understanding, being able to fine tune and customize my system. Of course I didn't go blindly through the tutorial but only did steps that I understood. (In case something goes wrong) Also I started to use (x)stow for everything.
Requiring four months seems as if it was a side project, which is of course totally fine, but for any main task it really should not take more than a few days, even on a semi-slow computer (well, depends on how slow).
Exactly my experience. Back in the day when all I knew was Windows and a couple of BSDs I wanted to get into Linux, because it seemed like it was going to become very relevant in the next decades. After years of slacking I thought LFS would probably be the ideal choice. Turned out it wasn't, at least for me. I ended up "learning Linux" with distros like Gentoo, Arch and later even Ubuntu, and by just replacing my production systems with Linux, which were still running on OpenBSD mostly. I have revisited LFS a couple of years ago and honestly didn't take much from it. I get the appeal, but it's not for everyone.
Bare LFS is not enough (I run my own): you need a kind of userland (above glibc) multi-version system which some kind of "atomic-ish" switching (always have a stable SSH running in case something goes wrong, better than unplugging the system disk and fix it on another computer).
For linux, same thing, with a kind of flip-flop-ing for updates/fix/etc.
The insight that gives you will probably scare you: the current "open source" stack is an abomination (the worst is the SDKs I think).
That's why we need _LEAN_ open source, and that includes the SDKs.
I promised myself not to work for any of those anymore :)
https://github.com/vasi/bootstrappable-distros
Linux from Scratch - https://news.ycombinator.com/item?id=46709727 - Jan 2026 (103 comments)
Linux from Scratch - https://news.ycombinator.com/item?id=41747966 - Oct 2024 (159 comments)
Beyond Linux from Scratch - https://news.ycombinator.com/item?id=39547118 - Feb 2024 (17 comments)
Linux from Scratch Version 12.0 - https://news.ycombinator.com/item?id=37648808 - Sept 2023 (28 comments)
Linux from Scratch - https://news.ycombinator.com/item?id=33734685 - Nov 2022 (9 comments)
Linux from Scratch - https://news.ycombinator.com/item?id=30496018 - Feb 2022 (96 comments)
Linux from Scratch - https://news.ycombinator.com/item?id=29949311 - Jan 2022 (9 comments)
Linux from Scratch with Training Wheels - https://news.ycombinator.com/item?id=28820602 - Oct 2021 (41 comments)
Linux from Scratch 10.0 - https://news.ycombinator.com/item?id=24350738 - Sept 2020 (49 comments)
Linux from Scratch - https://news.ycombinator.com/item?id=24238015 - Aug 2020 (86 comments)
Major Proposed Changes to Linux From Scratch - https://news.ycombinator.com/item?id=23787526 - July 2020 (93 comments)
Linux from Scratch - https://news.ycombinator.com/item?id=20168343 - June 2019 (15 comments)
Ask HN: Is the Linux From Scratch project still relevant? - https://news.ycombinator.com/item?id=20149111 - June 2019 (7 comments)
Linux from Scratch - https://news.ycombinator.com/item?id=16823110 - April 2018 (1 comment)
Linux from Scratch Version 8.2 released - https://news.ycombinator.com/item?id=16510333 - March 2018 (2 comments)
Linux from Scratch – build your own Linux distro - https://news.ycombinator.com/item?id=11829373 - June 2016 (57 comments)
Linux from Scratch - https://news.ycombinator.com/item?id=8392057 - Sept 2014 (1 comment)
Welcome to Linux From Scratch - https://news.ycombinator.com/item?id=4488162 - Sept 2012 (71 comments)
Linux From Scratch 7.1 Published - 3.2.6 Kernel + GCC 4.6.2 - https://news.ycombinator.com/item?id=3677350 - March 2012 (13 comments)
Linux From Scratch 7 Released - https://news.ycombinator.com/item?id=3171448 - Oct 2011 (27 comments)
Ask HN: Linux from Scrat...
Is there a Web version running in a VM/container in the browser?
I kind of need a LFS variant that is more flexible. For instance, the part up to having a working standalone compiler, should ideally become more robust and/or decoupled from the actual LFS build process. Once I have a sane toolchain, the rest is trivial, ruby scripts solve that part for me. But the build up towards that is not so reliable. Things also fluctuate so much in the build process; I still need perl for instance, and python wants to add more and more complexity into the build system as well, even though meson/ninja is a nice combo. For some reason, the build situation on Linux is not very elegant to word this nicely. Then we also "now you need LLVM and glslang and spirv" and so forth, and so on. The complexity always increases. To some extent I can understand why slackware crumbled. How can a single person, even with scripts, keep that all working?
It used to be a teenager’s rite of passage to build your kernel config, figure out LILO (remember that, pre-grub!), ipchains.. talk about feeling old!