Absolute Linux has reached the end – where to next?
- Reference: 1736854213
- News link: https://www.theregister.co.uk/2025/01/14/the_end_of_absolute_linux/
- Source link:
On December 15th 2024, developer Paul Sherman announced the end of the Absolute Linux project. This year, its website has already gone, but the Internet Archive's [1]final snapshot says:
End of the line for Absolute
Age, expense, but mostly LACK OF TIME leave me no choice but to give it up. I won't bore you with the crybaby details, but I gotta make ends meet. If someone wants to take the distro over, I would be happy to freely pass it on. I enjoyed tinkering all those years!
The project had been going since 2007, as the [2]oldest snapshot 's copyright message shows. Its [3]project page on Sourceforge is still there, complete with downloads, if you want to give it a try. Absolute Linux was a cut-down lightweight remix of the [4]venerable Slackware Linux , which at 2 5 years old (or 0x20, in old money) is the oldest surviving distro. As we said in that story, Slackware itself is anything but lightweight – it nearly filled our 16GB test root partition. That would seem to leave room for a lightweight remix, and we like the sound of some of its choices – the [5]nearly as venerable IceWM and a Reg FOSS desk favourite, the [6]ROX Filer .
We're always sad to see a project close down – although if any fans of Absolute Linux read this, it may be worth contacting Mr Sherman and offering to take it over. If that sounds over your pay grade but you want a lighter-weight Slackware, then [7]SLAX is still around , as are [8]Salix OS and [9]Zenwalk .
Which illustrates our core point here: there's a great deal of overlap between the many different distros out there, even if we restrict the criteria to, say, "cut-down versions of Slackware" – and there used to be [10]a dozen more .
[11]
And yet, some of the more radically experimental distros are sadly neglected, and there are established ideas in filesystems, packaging, desktop design and more that have yet to be adequately explored.
Directories and desktops
One of this vulture's favourites is [12]GoboLinux , which [13]as we've described before , discards the traditional Unix filesystem hierarchy and replaces it with a much simpler and more readable one modelled on macOS. Throwing out the hierarchy isn't unique: for instance, both [14]NixOS and [15]GNU Guix do that too. What their developers failed to do, though, was realise that replacing it with a hash-based machine-generated index does not make it appealing to most people, whatever the technological merits may be. We feel that [16]it must be easily human-readable as well .
Gobo uses the filesystem to keep programs, and all their libraries and dependencies, isolated in their own folders. This is also what macOS does with its [17].app bundles . The [18]GNUstep Desktop Environment , GSDE, uses the same .app format: both take the design from NeXTstep.
[19]
[20]
The ROX Filer that Absolute Linux used is part of a complete environment, the [21]ROX Desktop . ROX has its own form of application bundles, too – indeed, [22]AppImage uses the same structure, which is called the [23]AppDir format .
There is an obvious synergy between these directory-based apps and the GoboLinux design. A distro that combined the Gobo OS layout with either GNUstep and its app bundles, or ROX and its AppDirs – or both! – could work very well.
The 'minimize how much it sucks' approach
Since 2006, the [24]suckless project has been creating, in its own words, "software that sucks less". It even has a [25]page explaining its philosophy ; in brief,
Our philosophy is about keeping things simple, minimal and usable. We believe this should become the mainstream philosophy in the IT sector.
It's hard to fault that position, or indeed, that ambition.
The project's name is a nod to the computing proverb: [26]"all hardware sucks, all software sucks." Our interpretation is that project's name intentionally has a double meaning: by trying to make something which is "suckless" – that is without "suck" , if it were an abstract noun – the overall the project can help Linux to incrementally improve: to suck a little less . In other words, by trying to make software that's as good as it can possibly be, make the whole system less bad – you improve it. There's merit in the idea of striving toward perfection by optimization.
[27]
The suckless approach has led to some entire distributions with especially ruthless approaches to keeping things simple. [28]Sta.li is short for Static Linux – all its programs are statically linked: in other words, any library functions an executable needs are compiled into the binary.
The alternative, shared libraries, goes all the way back to [29]MULTICS in 1965 [PDF] – but it was standardized in UNIX later than you might expect: Sun [30]proposed a method [PDF] as late as 1987, and it was [31]still controversial in 1991.
Today, it's ubiquitous, but mainly in a theoretical way. Drew DeVault [32]tested this in 2020 , concluding:
Over half of your libraries are used by fewer than 0.1 percent of your executables.
Plan 9 from Bell Labs was the [33]successor to UNIX , and its designers [34]chose static linking instead of [35]dynamic linking .
When dynamic linking was proposed, part of the idea was that if a vulnerability was discovered and fixed, the system administrator would only have to update a single library, and all programs would thereby receive the fix. Of course, in real life, the result was [36]dependency hell , best known from its Windows form – [37]DLL hell .
[38]
In these days of [39]continuous integration and continuous delivery , recompiling everything to incorporate a fix sounds like standard operating procedure. Done right, it would barely be noticeable.
Stali wasn't the only attempt at an entirely statically-linked Linux distro; for instance, [40]Oasis Linux is still maintained. One extreme end result of this line of research is to link the entire OS into a single file, as done by both [41]Monolinux and [42]GaryOS . We suggest that something like this could fit well with the concept of [43]OneFileLinux .
Another ultra-minimalist distro that broadly followed Suckless principles, [44]KISS Linux , is officially dormant now, although the [45]user community are updating its [46]repositories . It wasn't completely statically-linked, but there has been [47]research in that direction .
Its creator Dylan Araps – whose best known work is surely [48]neofetch – retired, [49]saying : "Have taken up farming." We're reminded of our favorite burned-out engineer story, quoted in the [50]Pulitzer-winning The Soul of a New Machine : the message left on a terminal saying "I'm going to a commune in Vermont and will deal with no unit of time shorter than a season."
Find giants whose shoulders you can stand on
There is another approach to simplicity that's less extreme than the Suckless project's ruthlessness, but nonetheless productive: try to find the simplest existing alternatives, and assemble them in the simplest way.
This is one of the reasons that [51]we admire Alpine Linux , and we're not alone in this. Yes, millions of Docker users run Alpine, but most of them never see it, so they barely count except as testament to its fitness for use. One thing that struck us, though, was that Drew DeVault (yes, him again) wrote about it more than once: [52]in praise of Alpine Linux , and later, on why [53]Alpine Linux does not make the news . Perhaps the line that speaks to us the most is:
Alpine is the only Linux distribution that fits in my head.
For additional context, it's worth knowning that DeVault earlier [54]built his own distro , and more recently, [55]wrote his own Unix-like OS in his [56]own programming language . For us, this means his seal of approval on Alpine carries some weight.
Alpine takes many of the ideas from the dedicated miniaturized distributions designed for routers and other embedded roles, but expands and generalizes them enough so that you can use it as a general-purpose desktop or server. If you're willing and able to do the manual configuration, anyway.
Which seems to us to open up a whole area that's potentially ripe for exploration: a desktop-focused distro based on Alpine. (Just for clarity, neither [57]Adélie Linux nor [58]Chimera Linux are based on Alpine – they just share some components.)
An earlier lightweight distro on whose shutdown the Register reported [59]back in 2015 was Crunchbang Linux. That started out based on Ubuntu, then switched to Debian, before "Corenomial" threw in the towel. Since then, there's been both a community-led continuation, BunsenLabs, and [60]Ben Young 's one-man rebuild, CrunchBang++. When [61]we compared the two a year ago , we said that they remain quite similar and we'd love to see more difference, such as a systemd-free version based on Devuan.
But Devuan isn't significantly lighter-weight than Debian itself, which is a huge distro these days. As we [62]noted in "Drowning in Code" last year, Debian's [63]own materials say that version 12 contains "one and a third billion " lines of code. Neither BunsenLabs nor CrunchBang++ is massively smaller. The Debian derivative with the smallest footprint we've seen is the [64]x86-based Raspberry Pi Desktop . Sadly, it's not been updated since 2022 and it uses the now-dead Debian 11. All three of these distros are downloads of between 3-5 gigabytes, though, and the two CrunchBang derivatives take 4-6 gigs of disk and half a gig of RAM.
[65]Linus Torvalds offers to build guitar effects pedal for kernel developer
[66]Open source's totally non-secret weapon big tech dares not use: Staying relevant
[67]Sick of Windows but can't afford a Mac? Consult our cynic's guide to desktop Linux
[68]Top Linux distros drop fresh beats
With the same OpenBox window manager, Alpine uses half a gig of disk and about 90 MB of RAM. We would love to see someone combine Alpine with the CrunchBang config of [69]OpenBox and perhaps the [70]Calamares installer to create a seriously lightweight desktop Linux.
There is some precedent: the [71]ArchBang Linux was once a combination of the Crunchbang UI on top of Arch Linux. Over time, though, it has mutated, and now it's an Arch-based live medium with the Xfce desktop.
We feel that the most interesting space for innovation in Linux and distros is in the area of the smallest, simplest, lightest-weight distros. Leave the big corporate vendors to experiment with their [72]signed, sealed boot images and huge [73]cross-distro packaging systems .
As Richard Feynman said, [74]"There's plenty of room at the bottom" [PDF]. ®
Bootnote
There are many more experimental distributions out there than we had the space to mention here. One that looks quite interesting is [75]Glaucus , and we found maintainer Firas Khalil Khana's [76]"curated list of awesome projects" fascinating reading.
Get our [77]Tech Resources
[1] https://web.archive.org/web/20241229185520/https://www.absolutelinux.org/
[2] https://web.archive.org/web/20081211210544/http://www.absolutelinux.org/
[3] https://sourceforge.net/projects/absolute-linux/
[4] https://www.theregister.com/2023/07/20/slackware_turns_30/
[5] https://www.theregister.com/2022/10/05/icewm_version_3/
[6] https://rox.sourceforge.net/desktop/ROX-Filer
[7] https://www.theregister.com/2022/08/05/peppermint_os_and_slax_both/
[8] https://www.salixos.org/~
[9] http://www.zenwalk.org/
[10] https://en.wikipedia.org/wiki/List_of_Linux_distributions#Slackware-based
[11] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_software/oses&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=2&c=2Z4aYN0x1tDYrMVKhYc4xBwAAAQU&t=ct%3Dns%26unitnum%3D2%26raptor%3Dcondor%26pos%3Dtop%26test%3D0
[12] https://www.gobolinux.org/
[13] https://www.theregister.com/2021/12/03/nixos_linux_os_design/
[14] https://www.theregister.com/2022/12/13/nixos_2211_raccoon/
[15] https://www.theregister.com/2016/08/04/gnu_guix_emits_next_beta/
[16] https://www.theregister.com/2024/05/14/nix_forked_but_over_politics/
[17] https://developer.apple.com/library/archive/documentation/CoreFoundation/Conceptual/CFBundles/BundleTypes/BundleTypes.html
[18] https://onflapp.github.io/gs-desktop/index.html
[19] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_software/oses&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=4&c=44Z4aYN0x1tDYrMVKhYc4xBwAAAQU&t=ct%3Dns%26unitnum%3D4%26raptor%3Dfalcon%26pos%3Dmid%26test%3D0
[20] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_software/oses&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=3&c=33Z4aYN0x1tDYrMVKhYc4xBwAAAQU&t=ct%3Dns%26unitnum%3D3%26raptor%3Deagle%26pos%3Dmid%26test%3D0
[21] https://rox.sourceforge.net/desktop/
[22] https://appimage.org/
[23] https://docs.appimage.org/reference/appdir.html
[24] https://suckless.org/
[25] https://suckless.org/philosophy/
[26] http://www.catb.org/jargon/html/A/All-hardware-sucks--all-software-sucks-.html
[27] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_software/oses&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=4&c=44Z4aYN0x1tDYrMVKhYc4xBwAAAQU&t=ct%3Dns%26unitnum%3D4%26raptor%3Dfalcon%26pos%3Dmid%26test%3D0
[28] https://sta.li/
[29] https://homes.cs.washington.edu/~arvind/cs422/lectureNotes/dynlink-6.pdf
[30] https://www.cs.cornell.edu/courses/cs414/2001FA/sharedlib.pdf
[31] https://www.tuhs.org/Usenet/comp.unix.internals/1991-May/001441.html
[32] https://drewdevault.com/dynlib.html
[33] https://www.theregister.com/2024/02/21/successor_to_unix_plan_9/
[34] http://9p.io/wiki/plan9/why_static/index.html
[35] http://harmful.cat-v.org/software/dynamic-linking/
[36] https://en.wiktionary.org/wiki/dependency_hell
[37] https://en.wiktionary.org/wiki/DLL_hell
[38] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_software/oses&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=3&c=33Z4aYN0x1tDYrMVKhYc4xBwAAAQU&t=ct%3Dns%26unitnum%3D3%26raptor%3Deagle%26pos%3Dmid%26test%3D0
[39] https://www.theregister.com/2023/03/02/cicd_supply_chain_security/
[40] https://github.com/oasislinux/oasis
[41] https://github.com/eerimoq/monolinux
[42] https://garybgenett.net/projects/gary-os/
[43] https://www.theregister.com/2024/09/09/onefilelinux_esp_distro/
[44] https://kisslinux.org/
[45] https://kisscommunity.org/
[46] https://codeberg.org/kiss-community
[47] https://dilyn.cc/blog/KISS-static
[48] https://github.com/dylanaraps/neofetch
[49] https://github.com/dylanaraps/dylanaraps/blob/master/README.md
[50] https://www.pulitzer.org/winners/tracy-kidder
[51] https://www.theregister.com/2024/12/17/alpine_linux_321/
[52] https://drewdevault.com/2021/05/06/Praise-for-Alpine-Linux.html
[53] https://drewdevault.com/2023/07/25/Alpine-does-not-make-news.html
[54] https://drewdevault.com/2017/05/05/Building-a-real-Linux-distro.html
[55] https://drewdevault.com/2024/05/24/2024-05-24-Bunnix.html
[56] https://harelang.org/
[57] https://www.theregister.com/2024/12/20/adelie_linux_1_beta_6/
[58] https://www.theregister.com/2023/02/13/chimera_non_gnu_linux/
[59] https://www.theregister.com/2015/02/09/brit_linux_distro_crunchbang_calls_it_quits/
[60] https://github.com/computermouth
[61] https://www.theregister.com/2024/01/31/crunchbang_versus_bunsen_labs/
[62] https://www.theregister.com/2024/02/12/drowning_in_code/
[63] https://micronews.debian.org/2023/1686455026.html
[64] https://www.theregister.com/2022/07/11/raspberry_pi_desktop_update/
[65] https://www.theregister.com/2025/01/13/linus_torvalds_guitar_pedal_offer/
[66] https://www.theregister.com/2022/10/31/opinion_column_relevance_in_business/
[67] https://www.theregister.com/2022/05/31/the_cynics_guide_to_linux/
[68] https://www.theregister.com/2024/01/25/linux_mint_mx_updates/
[69] https://openbox.org/
[70] https://calamares.io/
[71] https://archbang.org/
[72] https://www.theregister.com/2022/10/26/tightening_linux_boot_process_microsoft_poettering/
[73] https://www.theregister.com/2023/06/09/will_flatpak_and_snap_replace/
[74] https://calteches.library.caltech.edu/1976/1/1960Bottom.pdf
[75] https://glaucuslinux.org/
[76] https://github.com/firasuke/awesome
[77] https://whitepapers.theregister.com/
Re: Debian can be tiny
[Author here]
> Debian can be tiny
Not very, no. I very specifically pointed out 3 of the smallest remixes of stock Debian that I know of: Crunchbang++, BunsenLabs, and the Raspberry Pi Desktop.
All are around 5GB of installed size, and the two OpenBox distros use about half a gigabyte of RAM, while the LXDE one uses about 200 MB.
Now, to me, that is not small, let alone "tiny". That's **BIG**. Really big. Vastly, hugely, mind-bogglingly big.
Then I compared to the PC version of Raspbian, which is the smallest full, general Debian remix I know of.
(Parenthetical disclaimer: yes, antiX is smaller, but it's a grab-bag of unrelated tools, while Raspbian is admirably very focussed. DSL 2024 is a smaller antiX but it still has multiple WMs, multiple file managers, multiple text editors, etc.)
This is still not small.
Then, I carefully compared this with Alpine Linux, noting that installing Alpine with OpenBox and the fonts etc. needed to get it working results in a disk footprint comparable to the Debian-based distros' _RAM_ footprint, and that the RAM footprint is under 1/4 of the other two and less than half that of the Pi Desktop.
Which is my attempt to falsify the claim that Debian is small or lightweight, _or can be made so_.
> Those images are about 620MB (compressed) to fit on a CD
Still no.
1. This is a false comparison. It fetches gigabytes off the internet while installing.
2. A full 2/3 of a GB should be enough for a full graphical desktop with internet tools. It was 20Y ago and the core tools are not that much bigger. Again: DSL manages it.
3. Who cares how big this installation image is? That's irrelevant. You only download it once. What you use every day is what matters.
> Later if you want a cut-down installation you can build one up that suits you from net-inst.
Still no. This takes some skill and expertise; that is the reason that things like Raspbian exist. Just "not picking so much stuff" is not a way to a lightweight system.
[Yes, out of order]
> what is the point of a cut down distribution?
Less code == faster. The less you load, the less you run, the quicker it goes.
Less code == safer. Smaller attack surface.
Less code == easier to understand. We need to remember this stuff, to navigate it. Simpler is better. Less is more.
Less code == lower system requirements. If you have an old slow computer, it works better. More and older computers work with it. If you have a new fast one, more of your expensive kit works for you on what you are doing and not on moving the bloat of the OS around.
Re: Debian can be tiny
“A full 2/3 of a GB should be enough for a full graphical desktop with internet tools. It was 20Y ago and the core tools are not that much bigger. Again: DSL manages it.“
Can I quote you as “640M ought to be enough for anybody”? ;)
Re: Debian can be tiny
I think we are talking a little at cross purposes. Here is some fragments of my background. Many years ago I created and used a personal distribution using [1]Linux from scratch . There are dozens of people on the planet who can benefit from understanding how to put a distribution together from source code. If nothing else, it gives you a huge appreciation of the effort required to maintain a distribution. I have put together minimal systems using [2]openwrt . These once valuable skills are steadily getting as useful as the ability to write software in assembly language because flash is cheap and memory is abundant.
I fully understand the value of a minimal _installation_ and I prefer light weight user interfaces for all the reasons you give for cut down _distributions_. Because of my background I can build such things out of Debian. You can start with a command line only system then if necessary install xorg plus one of the light weight window managers and end up with something small. It is not a task I would inflict on any moderately skilled user. One such GUI system works fine with 4GB of RAM and an unused swap partition. The OS takes up about 6GB but it includes compilers, developer libraries, video, image and document tools with a variety of pretty fonts.
A cut down distribution gets you to a similar system with far less need to do any research. The cost of that specialist distribution is you are dependent on a very small group of under appreciated maintainers to keep up with security updates. Building up from pieces of a mainstream distribution taps into a much larger group of over worked maintainers. My guess is that the maintainer to software ratio is better for the mainstream distributions.
For normal people a little extra money spent on hardware saves a significant amount of bother dealing with issues caused by trying to fit into a tight space.
[1] https://www.linuxfromscratch.org/
[2] https://openwrt.org/
Re: Debian can be tiny
There's also
Less code != doesn't include what I need
You could build a very small distro with a kernel, busybox and whatever additional libraries, if any, busybox needs. You're probably not going to do that.
You're probably not going to install the whole of Debian either.
The trouble is that your Goldilocks set of packages in between those extremes probably isn't going to be the same as mine and neither will be the same as the OP's nor anyone else's. The price of producing a distro that will satisfy a reasonable number of people is accepting that everyone will agree it contains cruft and nobody will agree on what that cruft is. The best we can hope for is that if we plot perceived amount of cruft against the number of people finding the useful we'll find that the curve flattens out.
I'm old enough to remember
When I could install and run some version of Redhat on a Libretto CT50. With a Pentium, 16MB of ram, and 800MB of disk - though to be fair, I did enlarge the disk rather quickly.
Re: I'm old enough to remember
My first home Red Hat box was a 486/66 with I think 8MB of RAM. It was a server, though, with no GUI.
Re: I'm old enough to remember
Happy Days. I just checked and I think my first distro was SuSE 5.2, late 90s, which came free on a magazine cover. Can't remember the hardware now, but as with every computer since there never seems to be enough memory or disk space - even today!
Not necessarily the best approach
One of the problems with any distro is getting all the dependencies you need for the functionality that you need, not necessarily all the dependencies that there can be (especially for lightweight systems) and definitely not too few to meet your requirements.
The second problem is the interface, which should be based around how you think and work and not by how the distros admins think and work. But, again, that creates interacting dependencies.
The third problem is the kernel, you really want a kernel configured to make the most of your system, not the dev's system.
Probably the simplest solution would be to not have a distro at all, at least not in the normal way of thinking, but rather have a requirements document you post plus a systems analysis, which goes to the distro provider. They then build a custom roll, configured to suit you, that still meets the provided test suites plus any distro test suites.
You then get that roll. It has what you need, and nothing more. Everything else the builder knows about can be installed as an extra, but what boots up first time is a distro that only has what you want.
The downsides of this are that updates and extensions take a lot longer to get to the user, a change in requirements can lead to a huge rebuild, validation is going to have to be a lot more basic, and the distro provider will need a powerful server farm, but it means you should get the best performance, the lowest footprint, and the least clutter.
Re: Not necessarily the best approach
And it will be expensive. Nobody's going to do that for every user for free. Not even Microsoft who does nothing for free, is going to do that for every user.
A more practical one would be something like a netinst iso which then presents you with a long menu of what you want to do and then install the appropriate options. An even more practical one is what Debian (and Devuan) do which is a netinst iso and a limited menu of bundles or, for convenience, if you're going to build more than one installation, an iso which comprises not only the netinst functionality but a lot of the addition S/W that's commonly downloaded.
Not for me anymore.
I admire the effort and resourcefulness of people developing this type of distro and twenty years ago I would have been straight in there fiddling about, tweaking and seeing how I could do more with less.
These days I just want a distro that "just works". I gave up PC wrangling a long time ago and the hobby side of Linux has faded away.
Horses for courses of course but with the vast amounts of storage and computing power now available I am content to let others push the envelope of Linux and settle for a larger, easier type of distro.
I do think that the distros under discussion here should be given a warning "Not for Windows refugees or beginners."
Re: Not for me anymore.
I have some experience setting up Alpine Linux. The installation and setup were really simple.
The only really tricky bit was getting the UEFI config right -- which has nothing to do with Alpine as such.
Re: Not for me anymore.
The only really tricky bit was getting the UEFI config right...
It's funny you should say that. I am currently having a nightmare trying to get a UEFI system to boot, Again nothing to do with the distro or the hardware, it's just as confusing as hell.
So far the distro's forum has been helpful but I am missing something. It's a pity as my new box has specs far in advance of my current one and I am getting frustrated with the whole thing.
There must be a reason for UEFI but for me it is just a PITA.
Thanks for this brilliant article. There are not a lot of publications and journalists exploring this subject so I’m always happy to discover a new post by yours truly.
Debian can be tiny
It has been a very long time since I tried a "complete" Debian install. It looks like the images are currently 4GB (compressed) to fit on a DVD. I have no idea how much that expands to if you install everything from the image. I am sure that is not anything like the whole of Debian but it should be plenty to get most people started.
I am more of a net-inst person. Those images are about 620MB (compressed) to fit on a CD. The intention is that installing most of the CD gives you a comfortable environment to download more as needed. More will almost certainly be needed and some knowledge of Debian will be required to get it.
I do not know if they are still around but there have been [1]business card sized Debian images. These fit into 50MB. They contain a sufficient subset of the command line environment that someone who knows what they are doing can download more and build it up to a a system tailored for a specific function - or all the way to something with far more software than any one person can use.
My point is, what is the point of a cut down distribution? Some poor maintainer has to dedicate his life to keeping it up to date and potential users have to install several distributions until they find one that is cut as needed. Much easier to install a complete distribution on a (modern) small SSD and find out over time what you actually use. Later if you want a cut-down installation you can build one up that suits you from net-inst.
[1] https://en.wikipedia.org/wiki/Bootable_business_card