Timeline
Post
Remote status
Context
21
@lanodan@queer.hacktivis.me @volpeon@icy.wyvern.rip Just because Arch is terrible doesn't mean this applies to every other distro though.
@SuperDicq@minidisc.tokyo @lanodan@queer.hacktivis.me @volpeon@icy.wyvern.rip it does not have to do anything with arch itself. This can happen with every distro, if GRUB gets an update but fails or if the initcpio-Image was not build successfully.
The best way to avoid those issues is, by reading through the output of your package manager and ensuring that the installation of GRUB and the Initcpio process were successful.
@kitsune_yasu@waldbewohner.eu @lanodan@queer.hacktivis.me @volpeon@icy.wyvern.rip
This can happen with every distro, if GRUB gets an update but fails or if the initcpio-Image was not build successfully.I'm not that familiar with Arch, but it sounds extremely incompetent to me that Pacman would just continue updating and not fail if grub-install fails.
Debian, for example, has a transaction based package manager so that makes sure when failures like this happen the old stuff that was previously working is not overwritten. So that definitely sounds like a design flaw in Pacman.
@SuperDicq @lanodan @kitsune_yasu @volpeon pacman never does anything with a bootloader. There are two (three) main failure modes that pacman has:
- distro keyring not up to date, failing package authenticity check (archlinux-keyring needs to be updated first)
- pacman deletes initrds on transaction start meaning that if your system fails in the middle of an update, you not only have a system in an unknown state with half-extracted packages, but also no kernel to boot
- when using BTRFS, the free space check can fail because free space as reported by BTRFS isn't actually real and it doesn't use the specific btrfs library for getting free space
- distro keyring not up to date, failing package authenticity check (archlinux-keyring needs to be updated first)
- pacman deletes initrds on transaction start meaning that if your system fails in the middle of an update, you not only have a system in an unknown state with half-extracted packages, but also no kernel to boot
- when using BTRFS, the free space check can fail because free space as reported by BTRFS isn't actually real and it doesn't use the specific btrfs library for getting free space
@phnt@fluffytail.org @SuperDicq@minidisc.tokyo @lanodan@queer.hacktivis.me @kitsune_yasu@waldbewohner.eu @volpeon@icy.wyvern.rip
- distro keyring not update to date failing package authenticity check (archlinux-keyring needs to be updated first)Wait, that's not supposed to happen?
@lolitechengineer @phnt @SuperDicq It's not, other distros just have keyring put into a package, so it gets updated normally.
I got that one in my containers and well… tossed Arch out definitively because there's a point where it's just pure incompetence.
I got that one in my containers and well… tossed Arch out definitively because there's a point where it's just pure incompetence.
@lanodan @lolitechengineer @SuperDicq It is packaged and it gets updated when you update the system, but when you don't update for a long time, new packages can be signed with keys that aren't in your old system keyring and there's no automatic pre-upgrade step for updating that keyring. Hence why the manual update is necessary.
Gentoo solved this issue by force-updating the wkd on every update before emerge even resolves dependencies.
Gentoo solved this issue by force-updating the wkd on every update before emerge even resolves dependencies.
@phnt@fluffytail.org @lanodan@queer.hacktivis.me @lolitechengineer@loli.church At least we can all agree here that Arch is terribly designed.
@SuperDicq @lanodan @lolitechengineer Nope, still the best power user distro if you don't need the use flag autism of Gentoo. Everything else including Debian is a bad joke (except Mint for newbies).
@phnt@fluffytail.org @lanodan@queer.hacktivis.me @lolitechengineer@loli.church Also the best power user distro is GNU Guix.
How can you even call yourself a power user if you have never learned the power of Scheme?
@SuperDicq @lanodan @lolitechengineer ((((((((((lisp))))))))))
I'm not touching that with a ten foot pole
I'm not touching that with a ten foot pole
@phnt@fluffytail.org @lanodan@queer.hacktivis.me @lolitechengineer@loli.church That's fine, not everybody has to be a power user.
@SuperDicq @lanodan @lolitechengineer It is gnuware anyway, so I wouldn't use it either way unless I absolutely had to.
@phnt@fluffytail.org @lanodan@queer.hacktivis.me @lolitechengineer@loli.church You one of those weirdo Alpine/Chimera GNU avoiders? Having nightmares of Stallman's feet?
@SuperDicq @lanodan @lolitechengineer No, but I would run LLVM/musl Gentoo profile if I didn't want to use Steam.
@SuperDicq @lanodan @lolitechengineer Not for technical reasons (except glibc sucking), but just to spite people like you. :)
@shinyoukai@izu.laidback.moe @phnt@fluffytail.org @SuperDicq@minidisc.tokyo @lanodan@queer.hacktivis.me BSD seems really interesting but it seems its not ready yet
@lolitechengineer also Hyperbola is a fucking joke, 2026 and they never released @SuperDicq @lanodan @phnt
@shinyoukai@izu.laidback.moe @lolitechengineer@loli.church @lanodan@queer.hacktivis.me @phnt@fluffytail.org I personally think Hyperbola is a bit of a nutjob project in my opinion.
For example the only reason they are making a BSD version now I think is because they think Rust in Linux is bad. I mean I don't like Rust either, but is it that really enough of a reason to switch to fucking BSD? Wouldn't it be less bad to just remove the Rust parts from Linux?
Hyperbola also does not allow any software that is developed by a for-profit company. Yes this means that Hyperbola does not support basic shit like zstd because it is developed Facebook.
It's bonkers.
If I really wanted to screw up my system using pacman I'd rather install Parabola which seems like a project ran by sane people.
Replies
39
@SuperDicq @phnt @lolitechengineer @shinyoukai Yeah, it gives me vibes that are kind of similar to what you can find in suckless.org and related, lot of rambly declarations that aren't really linked to acts (similar to vaporware and virtue signals), barely anything to show and tends to define itself almost entirely as anti-
@SuperDicq @lolitechengineer @phnt @shinyoukai In a rather ironic way, almost the opposite of OpenBSD, which has done lot of things and tends to do reimplementations or redesigns when what's available doesn't works well enough for them (OpenSSH, LibreSSL, doas, openrsync, signify, …) instead of just throwing things away.
@lanodan @shinyoukai @lolitechengineer @SuperDicq At least with suckless you get what you signed up for. A base piece of software you patch into something that suits specifically you with the provided patches or your own. If you come to them expecting something usable out of the box, you will be disappointed no matter what. Or if you want a piece of software that is maintained but never changes.
With Hyperbola it's removing things for the sake of removing things, because we don't like some abstract interpretation of dumb politics. In a way, it perfectly fits the GNU maximalism mindset. Instead of letting a user choose whether they want to use non-free blobs on _their own_ hardware, we choose for them and remove that ability to even do so, because we are smarter than the user, and because we hate choice, and letting the user patch our patches so that they can do what they really want. Just because we have some abstract dumb opinion on what proprietary firmware is. There's no actual progress made, just removal of options for the sake of
removing them, while thinking you are holier than thou.
And then if you come to the BSD crowd, it is a completely different mindset. Instead of coming to a discussion being dead set on the "only proper way" and arguing only for that, the discussion starts level headed and without predetermined outcome.
With Hyperbola it's removing things for the sake of removing things, because we don't like some abstract interpretation of dumb politics. In a way, it perfectly fits the GNU maximalism mindset. Instead of letting a user choose whether they want to use non-free blobs on _their own_ hardware, we choose for them and remove that ability to even do so, because we are smarter than the user, and because we hate choice, and letting the user patch our patches so that they can do what they really want. Just because we have some abstract dumb opinion on what proprietary firmware is. There's no actual progress made, just removal of options for the sake of
removing them, while thinking you are holier than thou.
And then if you come to the BSD crowd, it is a completely different mindset. Instead of coming to a discussion being dead set on the "only proper way" and arguing only for that, the discussion starts level headed and without predetermined outcome.
@lanodan @SuperDicq @lolitechengineer @shinyoukai Btw if anybody wants to find out where the "only proper way" mindset in GNU and related software comes from, look no further than at mailing lists when RMS was still involved in development.
@phnt@fluffytail.org @lanodan@queer.hacktivis.me @lolitechengineer@loli.church @shinyoukai@izu.laidback.moe >Tries to following up the bait with more bait hoping we will bite.
@SuperDicq @lanodan @lolitechengineer @shinyoukai That's the thing, I'm not baiting. The holier than thou mentality is real. You don't see it, because you mostly agree with it. If I wanted to bait, I wouldn't write multiple paragraphs explaining my thinking process.
@phnt @shinyoukai @lolitechengineer @SuperDicq Suckless did some software that can work, but their community has those kinds of weird "Remove things just because we don't like it, keeping things in a working state? Naaah".
Like take a look at stali, or almost any other suckless distro, there's typically no C compiler at all, and a bunch of suckless software (dwm, dmenu, …) are entirely absent because well no graphical session.
(One exception is Oasis Linux, but not sure how suckless related it is, mcf might just have grabbed sbase because it's simple to compile)
Like take a look at stali, or almost any other suckless distro, there's typically no C compiler at all, and a bunch of suckless software (dwm, dmenu, …) are entirely absent because well no graphical session.
(One exception is Oasis Linux, but not sure how suckless related it is, mcf might just have grabbed sbase because it's simple to compile)
@lanodan @shinyoukai @lolitechengineer @SuperDicq The community is kind of exactly what you expect from wannabe power users that want to feel better than everybody else. A very reddit-esque behavior. They are basically larpers that call everything they don't want or agree with bloat like the meme that died at decade ago.
When you remove the community from the project and its maintainers, you get a pretty high quality pieces of software (except dumb variable names) ready to be hacked on. But that is the case with many projects these days. Look at a certain oxidized language that was interesting and ruined by the annoying community around it.
>suckless distros
Never heard of them and making a suckless distro is the dumbest thing you can do with it. Since the software is made to be patched by a user directly and not properly packaged. The only things that properly work as expected without patching is dmenu, maybe dwm and probably the init system.
When you remove the community from the project and its maintainers, you get a pretty high quality pieces of software (except dumb variable names) ready to be hacked on. But that is the case with many projects these days. Look at a certain oxidized language that was interesting and ruined by the annoying community around it.
>suckless distros
Never heard of them and making a suckless distro is the dumbest thing you can do with it. Since the software is made to be patched by a user directly and not properly packaged. The only things that properly work as expected without patching is dmenu, maybe dwm and probably the init system.
@phnt @shinyoukai @lolitechengineer @SuperDicq > Since the software is made to be patched by a user directly and not properly packaged.
It could be like AT&T Unix/BSD/Plan9, where you go in /src or similar directory, do your edits, and type a command like make to recompile, but for this it needs to be a non-toy system where you have things like development tools.
@jesu @shinyoukai @lolitechengineer @SuperDicq @lanodan Autistic about POSIX compliance. Their gcc build doesn't install the c89 and c99 wrappers, so build systems can break.
@jesu @phnt @shinyoukai @lolitechengineer @SuperDicq Specially with POSIX mention, like hopefully tossing out c89/c99/c17 (each version requires a single wrapper) is the only thing it explicitly does.
(I also have those wrappers tossed out on my system but to be sure cross-compiling will work)
(I also have those wrappers tossed out on my system but to be sure cross-compiling will work)
@lanodan @shinyoukai @lolitechengineer @SuperDicq @jesu There probably is some reasoning behind that I never bothered to check, but it is a bit weird choice.
@lanodan@queer.hacktivis.me @phnt@fluffytail.org @shinyoukai@izu.laidback.moe @lolitechengineer@loli.church I'm not a fan of suckless software. It seems very esoteric and pointless to have minimalism as your end goal.
You can just write software that performs well AND has a lot of features. I find their philosophy flawed.
@SuperDicq @phnt @lolitechengineer @shinyoukai Well minimalism can be important but it shouldn't be the primary objective and I think that's the point that suckless didn't get, plus well… minimalism to me is something that you do not just in terms of code but also dependencies with a bit of preference for OS-scale from time to time.
OS scale because I think few things in modern systems should be split in more discrete components, specially as corporate mindset has a tendency to centralise policies into huge monolithic software that doesn't cooperates with the rest of the system in an equal manner.
OS scale because I think few things in modern systems should be split in more discrete components, specially as corporate mindset has a tendency to centralise policies into huge monolithic software that doesn't cooperates with the rest of the system in an equal manner.
@lanodan@queer.hacktivis.me @phnt@fluffytail.org @lolitechengineer@loli.church @shinyoukai@izu.laidback.moe I'm personally in favor of decentralization too. I do not like huge monoliths either.
But I am not in favor of minimalism for the sake of it. Make more packages that do more things. I like having many features that make computing easier and more convenient.
@SuperDicq @phnt @lolitechengineer @shinyoukai Yeah, I think minimalism shouldn't be about removing overall features, more about refactoring and keeping things clean, which are very secondary but are effectively needed long-term.
Like you don't throw away everything in your home on a regular basis or should live like diogene, but getting some stuff away is needed once in a while.
Like you don't throw away everything in your home on a regular basis or should live like diogene, but getting some stuff away is needed once in a while.
@lanodan@queer.hacktivis.me @phnt@fluffytail.org @lolitechengineer@loli.church @shinyoukai@izu.laidback.moe For example, my fully refactored version of Polygon Love 2 contains more lines of code than the original project did. Even though I actually removed some unused features from the code.
The original project was a giant spaghetti mess. The entire game was contained in only 10 source code files, most of them consisting of giant 3000 line god classes. I split all of those into actual modules and stuff like that so the project is actually readable and maintainable now.
@lanodan@queer.hacktivis.me @phnt@fluffytail.org @lolitechengineer@loli.church @shinyoukai@izu.laidback.moe Actually now that I think of it most of the line count increase comes from giant lists of constants that originally were all compressed into one line. And also GPLv3 license header on every file.
@SuperDicq @phnt @lolitechengineer @shinyoukai > And also GPLv3 license header on every file.
Heh, you know about SPDX headers (and REUSE), right?
And well lines of code have pretty much always been useless and are a metric I love to hate.
Metrics like CPU/memory/… usage make more sense, but are very rarely used even though that's where ressources are scarce and can even make the difference between something which runs and something which doesn't.
Also reminds me that minimalism sometimes gives you two kinds of extremists:
- almost 0 dependencies
- dependency over own module (low line of code kind tends to go there)
Except that can end up just shifting where the complexity is, waterbed theory style.
Heh, you know about SPDX headers (and REUSE), right?
And well lines of code have pretty much always been useless and are a metric I love to hate.
Metrics like CPU/memory/… usage make more sense, but are very rarely used even though that's where ressources are scarce and can even make the difference between something which runs and something which doesn't.
Also reminds me that minimalism sometimes gives you two kinds of extremists:
- almost 0 dependencies
- dependency over own module (low line of code kind tends to go there)
Except that can end up just shifting where the complexity is, waterbed theory style.
@lanodan@queer.hacktivis.me @phnt@fluffytail.org @lolitechengineer@loli.church @shinyoukai@izu.laidback.moe I know about SPDX headers and REUSE obviously, but I personally prefer using the full header for the following reasons;
Unless you are already extremely familiar with software development something like # SPDX-License-Identifier: GPLv3-or-later is not going to mean anything to you. Especially considering that I am making a video game you have to assume that whoever stumbles onto this project might not be the most professional programmer.
The full license header actually explains very briefly what free software is, what the GPLv3 does and where to read more about it (the gnu.org website).
So I think it's also good promotion for the free software movement and its ideals to have full license headers in your projects prominently displayed.
@SuperDicq @phnt @lolitechengineer @shinyoukai Yeah, although in all files?
At least for me SPDX is enough for files and then could put something around ethics and copyright in something like a section of the readme (which the file header can refer to).
Which also has the effect of being visible beyond just the devs who happen to be reading the code.
At least for me SPDX is enough for files and then could put something around ethics and copyright in something like a section of the readme (which the file header can refer to).
Which also has the effect of being visible beyond just the devs who happen to be reading the code.
@lanodan@queer.hacktivis.me @phnt@fluffytail.org @lolitechengineer@loli.church @shinyoukai@izu.laidback.moe Of course the license is also referred to in the readme file.
But the reason you put a license header in every file is so that when a file gets separated from its project the context of its license is not lost.
This happens very often by the way. That someone just strips out a single module or a single file from an existing project and uses it in another project. In these cases the full license header is the most important.
@SuperDicq @lanodan @phnt @lolitechengineer @shinyoukai when I do that the license header will get chopped alongside the half of the file I'm not using
@latein @shinyoukai @lolitechengineer @SuperDicq @lanodan Copyright dates header in files is especially dumb, because you either have a huge commit bumping it every year while nothing changed. Or you forget to update it whenever you update the file.
@phnt @latein @shinyoukai @lolitechengineer @SuperDicq @lanodan The copyright date(s) being accurate is highly important - otherwise the copyright notice isn't valid.
If nothing changed in that file, the copyright year(s) should not be changed for that file.
For a very trivial change, the old copyright year is correct - for changes that are added a year later, that qualify for copyright, the new year should be added so the copyright header remains correct (GNU Emacs can do this automatically for you).
If nothing changed in that file, the copyright year(s) should not be changed for that file.
For a very trivial change, the old copyright year is correct - for changes that are added a year later, that qualify for copyright, the new year should be added so the copyright header remains correct (GNU Emacs can do this automatically for you).
@Suiseiseki @latein @shinyoukai @lolitechengineer @SuperDicq @lanodan
>otherwise the copyright notice isn't valid.
Everything should be public domain anyway. Licensing is stupid. Copyright is stupid. If you care more about how your software is distributed and licensed than when someone uses it, you can't be helped. You are a lost cause.
fedi-fsf-gnu-anniversary.png
>otherwise the copyright notice isn't valid.
Everything should be public domain anyway. Licensing is stupid. Copyright is stupid. If you care more about how your software is distributed and licensed than when someone uses it, you can't be helped. You are a lost cause.
fedi-fsf-gnu-anniversary.png
@phnt @latein @shinyoukai @lolitechengineer @SuperDicq @lanodan >Everything should be public domain anyway.
Hurry up and overthrow the government then - that is the only way to achieve that.
>If you care more about how your software is distributed and licensed than when someone uses it
If the software is not competently licensed and distributed and therefore not is legal to use for any purpose, then people cannot use it.
Hurry up and overthrow the government then - that is the only way to achieve that.
>If you care more about how your software is distributed and licensed than when someone uses it
If the software is not competently licensed and distributed and therefore not is legal to use for any purpose, then people cannot use it.
@phnt@fluffytail.org @Suiseiseki@freesoftwareextremist.com @latein@cawfee.club @shinyoukai@izu.laidback.moe @lolitechengineer@loli.church @lanodan@queer.hacktivis.me The only thing I care about is that my software will not become proprietary. Anything else doesn't matter.
@phnt@fluffytail.org @Suiseiseki@freesoftwareextremist.com @latein@cawfee.club @shinyoukai@izu.laidback.moe @SuperDicq@minidisc.tokyo @lanodan@queer.hacktivis.me well, you can tell whoever represents you in whatever legislature has jurisdiction over you that you think they should change the law. But that is not the case currently.
@phnt @shinyoukai @lolitechengineer @SuperDicq Yeah, can just put no dates at all or something like the year of the start of the project.
(And fsfe's reuse acknowledges that)
This stuff always reminds me how meanwhile Hololive just puts "ⓒ COVER" or "ⓒ 2016 Cover Corp." (even as of few months ago) on their official arts.
If a more or less showbiz agency can do this, surely most of foss projects can.
(And fsfe's reuse acknowledges that)
This stuff always reminds me how meanwhile Hololive just puts "ⓒ COVER" or "ⓒ 2016 Cover Corp." (even as of few months ago) on their official arts.
If a more or less showbiz agency can do this, surely most of foss projects can.
@lanodan@queer.hacktivis.me @phnt@fluffytail.org @shinyoukai@izu.laidback.moe @lolitechengineer@loli.church Yes, but you can get away with just mentioning the start date of the project but I do believe is better to mention both the start and last modified date. Being more clear is always better.
@phnt @shinyoukai @lolitechengineer @SuperDicq @lanodan I stopped having them for that reason, because when I am editing the file, I don't care about its header.
a sensible place for the license is the project root and the modifications from the timestamp / git log.
a sensible place for the license is the project root and the modifications from the timestamp / git log.
@phnt @shinyoukai @lolitechengineer @SuperDicq @lanodan GNU is the only OS that actually gives the user the choice of doing what they want to do.
Not attacking the user with proprietary software rather allows for choice (as the user can choose to have freedom by choosing not to throw it away), rather than taking away choices.
Every single BSD is proprietary software, thus there is no choice to have freedom - as you'll never have it.
Not attacking the user with proprietary software rather allows for choice (as the user can choose to have freedom by choosing not to throw it away), rather than taking away choices.
Every single BSD is proprietary software, thus there is no choice to have freedom - as you'll never have it.
@SuperDicq@minidisc.tokyo @shinyoukai@izu.laidback.moe @lanodan@queer.hacktivis.me @phnt@fluffytail.org
Hyperbola also does not allow any software that is developed by a for-profit company. Yes this means that Hyperbola does not support basic shit like zstd because it is developed Facebook.reminds me of their prohibition of VP8 (and webp), VP9, AV1, even though they are freely licensed formats with libre reference and community built encoders and decoders and instead suggest using... HEVC?
@SuperDicq @lanodan @phnt @lolitechengineer @shinyoukai Hyperbola's wiki + removed/hold list is some funny shit to laugh at, holding gimp to 2.10 because only want GTK2 version, git due to Rust, and mpv for some vague reason. Vaguely looking at the forum activity there's barely any users, and I only know off a Youtuber intentially installing it for his Thinkpad x60 last year. Might as well go Debian/Devuan and just don't turn on nonfree packages/firmware. It's fascinating reading those updates + wiki rambles from Hyperbola devs.
@lazyisolation@stereophonic.space @lanodan@queer.hacktivis.me @phnt@fluffytail.org @lolitechengineer@loli.church @shinyoukai@izu.laidback.moe Yes, Hyperbola is very opinionated for some very specific people who are also the only users.
If you want a good Arch-based distro with proprietary software removed you should install Parabola instead.