Years ago, as a kid, when the only computers I had access to were cobbled together hand-me-downs and machines people would throw away I used to do this kind of thing. Like the author, I settled on IceWM after trying Fluxbox/JWM/FVWM/"Window Maker" and all the various forks with minor variations.
It was fun to tinker, but it's questionable if I learnt anything useful other then how to get hobby DEs working on ancient hardware. I guess tangentially I gained an intermediate knowledge of Linux command line tools and some networking concepts.
PuppyLinux [0] ended up being my choice for easy defaults after DSL [1] stagnated. I'm glad to see it's still going and has developed its own ecosystem. Even after I could buy reasonable computers, for a while I took some pride in squeezing performance out of budget computers but ultimately, now that I need to be productive to get paid, it's nice to have computers that "just work" with modern software.
You also gained a comfort level poking around in moderately complex system that could break if you did something wrong, and then force you to step through problem solving to fix it. This is a very important skill. (Though it's also important to understand when potentially breaking something, even temporarily, is low stakes vs. something to be avoided at all costs)
It's a skill I like to call a "video game mentality" that I think was probably influential in the rise of technology: video games gave players the ability to push buttons and experiment in all kinds of ways with a technological system but in a very low stakes environment.
I don't think I would have been nearly as comfortable poking around and tweaking system variables in my first computers running Windows 3.X and 95 if I didn't have years of experience poking around in other pixel-based environments in the form of videogames. I would break things, panic, realize that whatever it was, it was fixable, go through trial and error, learn to find other people and resources to help, fix the problem, and do it all over again at some point.
I had the exact same experience, but a fairly clear flow into my career. My own tinkering led to a job at a local computer store fixing computers when I was about 15, and then a job as a build engineer at 18. This was in the 90s so there weren’t a ton of college students with computer experience to compete with, and I think there was more openness to non-traditional candidates.
My last job was product director for a major CI/CD platform, so there’s a pretty clear breadcrumb trail all the way back to tinkering with Linux on old hardware to now.
The Linux command line and scripting was part of it, but the real deeper lesson was that computers are understandable and if you are persistent you can fix or improve nearly anything, or assemble something new and useful by building on what’s available to you.
>Years ago, as a kid, when the only computers I had access to were cobbled together hand-me-downs and machines people would throw away I used to do this kind of thing. Like the author, I settled on IceWM after trying Fluxbox/JWM/FVWM/"Window Maker" and all the various forks with minor variations.
I had one of those original EeePC's years ago, 900 mhz celeron in like 2008, 800x480 screen, just woefully out of date. Ended up running Fluxbox just because of ram usage. With an actual WM it couldn't stream The Daily Show smoothly. I really do love playing with low spec systems.
> it's questionable if I learnt anything useful other then how to get hobby DEs working on ancient hardware
I did similar things, and on one hand I feel it’s ok that it’s not all “useful” - it was a hobby after all, but on the other hand I do feel it had benefits - you learned patience, creative thinking, problem solving, and working on those projects for a long stretch of time almost has a meditative effect I felt. I’m sorry that a lot of modern kids probably don’t get to experience this as much as everything is pre-packaged, system-on-a-chip, non-fixable etc. (And people are more careful of how they dispose of computers due to privacy).
Wow, Damn Small Linux, what a throwback! That was how I was first ever able to run and interact with a Linux system back as a teen. I had no computer with Administrator access, but I found a zip file with DSL, a QUEMU.exe, and a .bat file which would run QUEMU and boot DSL, cause I couldn't figure out how. I never even got it to save to a persistent disk, I had to boot and re-do all the changes I'd figured out the prior time. What an amazingly educational little thing that was!
Wow, this took me down memory lane. Puppy Linux and its crazed Windows 95-esque dialogue boxes (remember Puppy Unleashed?) will forever hold a soft spot in my heart.
Not sure how it is anymore, but Puppy Linux circa 2008 was like the GeoCities of operating systems, glorious in its own unpolished way. Vector Linux, too, which I believe was Puppy's ancestor.
> Years ago, as a kid, when the only computers I had access to were cobbled together hand-me-downs and machines people would throw away I used to do this kind of thing.
That was me well into my 20s :) Win2k broke, and I was broke, so the only option was to install linux.
It paved the way for me to get into programming, not so much shell scripts, but just the powerful idea that computer systems were malleable things you could change - something not really obvious for those of us who cut our teeth on MS Windows.
> it's nice to have computers that "just work" with modern software
Yeah I never understood people who work with computers all day economizing on their computers. Even from scalpers you’ll pay ~$6500 for top of the line parts and peripherals, which is about the cost of a high school student’s violin.
Yes, but it's also thanks to this mentality that a simple text editor occupies 1 gig of ram and two cpu cores nowadays. If the dev's machine is infinitely fast and has storage as fast as other people's ram, they stop caring about performance issues, because they never encounter them in the first place. And I'm not talking about working on hand written assembly for a week to get the last bit of performance. I've repeatedly found silly things like O(n) code calculating something that could be expressed as O(1), hilariously complicated xpath expressions that in the end just retrieved an immediate child and whatnot.
At one of my first jobs, when my boss got annoyed by some software being slow, je always said that developers should get the fastest hardware available, but as soon as they start testing or using their own software, the machine needs to magically turn into a 386 with your home directory on an NFS share. Granted, that was 20 years ago so 386 didn't sound as hilarious as it does now, but it was still a little extreme. But that idea stuck with me, and to this day I do test software on slow machines every now and then, looking for obvious performance left on the table.
I use a Raspberry Pi as my daily driver. I like it because the hardware is cheap and easy to replace. It sips electricity, which matters to me. It doesn't support Intel or AMD. And as a developer, I can more acutely feel differences in performance based on the algorithms I choose. If I can make it run acceptably on a Pi, it should work okay on a non-flagship phone or a 4+ year old office PC.
Wow that’s pricy. I played 1st section violin through college and only have a $2k one. My friend played first chair 1st violin section through college and had a $4k violin and it sounded amazing. Maybe we got lucky and found a really good instrument shop? Another friend had a cello that costed well over $15k though so there’s that.
At least you're engaging with the premise. You could also get a $980 with tax M1 MacBook Air. A Dell prebuilt - even the Alienware, which does not have to pay scalper prices and comes with a 3070 - is $2,611. These are all comparable options. In the instrument world, a $6,500 computer is more like a $30,000-150,000 violin and bow. Like the stuff professionals use.
1. In many countries $6500 is enough to start a business. I'd rather do that than pay for an overpriced piece of aluminum with a piece of fruit etched on one of the sides. What if I'm not the business type? That's fine. It's still probably better to save that money.
2. Modern computing is turning into a dumpster fire. It will be pretty soon that the Intel platform will work in the same way as the above-mentioned fruitty platform and you won't be able to run whatever software you want on it. While I agree OP is probably making his life extra miserable [1], I admire people like him, since their efforts will give us the next free/fun computing platform.
[1] I too run a Raspberry Pi at home, but I've long given up on turning it into a desktop machine. There is an easier way to run modern software on cheap hardware with a fraction of the hassle.
Easy, not everyone lives in countries where giving ~$6500 is a possibility, and if they have access to such money they wouldn't be spending it on computer parts most likely.
>KDE is my desktop environment of choice. KDE5 is rock-solid, configurable in any way possible and works great. It treats you like a responsible adult instead of a child like GNOME does these days, and after XFCE switched to GTK3, the RAM usage is on-par, more often than not a bare KDE install (Debian or Arch) uses around 300MB ram. This is with Baloo (search indexer) and Akonadi (PIM database backend) disabled.
The GNOME Foundation's supposed rationale for removing basic functionality from its desktop is to help inexperienced users, and international users. However, those are precisely the same users who are more likely to have a shortage of computing resources. GTK's bloat is really indefensible.
I might be mistaken here, but it seems to me that the international users who want a simple experience are already well served by smartphones (android or otherwise).
The international users who are going to install linux on a laptop are the tinkerers and fiddlers and the self motivated learners who aren't going to be motivated by a dumbed down, inaccessible interface.
It's important to remember that GNOME is primarily funded/developed by Red Hat, who are owned by IBM. Red Hat's goal isn't to make something for the kind of person who would install Linux on a laptop themselves -- they're trying to make a clone of macOS they can sell in bulk to enterprise customers. The intended user population is very different.
I'm not sure where you heard any of that, but I have never heard any Red Hat employees say anything to that effect, ever. Fedora still exists, and to me that seems perfectly fine to install on a random laptop.
Not necessarily with that line of thought, Red-Hat has already acknowledged many years ago that there is no money to be made on the GNU/Linux desktop (Mandrake, SuSE and Canonical also found out years later).
If Fedora (or GNOME, for that matter) fits your needs, you should feel free to use it! But if you're wondering, like the grandparent comments, why projects like GNOME seem to be moving away from fitting your needs, the answer is that they're being designed for Red Hat's enterprise customers first and any use you get out of them is a happy afterthought.
Again I'm not sure where you got that, I have not heard any Red Hat or Fedora developers say that's the focus. In fact if you follow GNOME development, there are various other directions that people are pulling in. It might be the case coincidentally for some things, but that could be yet another one of those happy afterthoughts.
Except GNOME rewrote GNOME Shell in JavaScript. I don't think GTK3 requires GJS, so I'm not sure why it uses more RAM than GTK2, maybe due to CSS? I know it's still Cairo-based, not GPU-accelerated or DOM-like.
I can't say for sure what the deal is with GTK3, but GTK4 apps are GPU accelerated and should be faster. Though it may come at a cost of using more GPU memory, depending on the app.
Yeah but if you don't like it, you can just leave it alone and it's perfectly functional. That said I use Pop_OS! which is gnome based and fine. I like the defaults just fine. I have used all sorts of DE over the years, in the end I decided I didn't really care :) . KDE is fine, Gnome is fine, xmonad is fine.
I've been adopting this mindset myself. I've used almost every prominent DE and WM there is, and they're all fine. Varying levels of ugly or beautiful perhaps, but I just want to get past them and onto my actual work.
GTK3/4 alone not really, as my Linux netbook is doing just fine with XFCE.
Now having GNOME with its Javascript based shell and extensions, making every click take a couple of seconds longer and me wondering what it is doing, no thanks.
The javascript is used mostly to drive the shell GUI, it shouldn't be taking seconds off all your mouse clicks. If there is really bad slowness, you should consider reporting that as a bug, it may not be the javascript that's at fault. The performance shouldn't be any worse than a typical lightweight web app running in firefox. That is of course if you ever end up using GNOME again for whatever reason.
I don't understand why you have to wonder what it's doing, the javascript is all self-contained and hosted locally. It's not using npm or anything like that. And the extensions are pretty much the same as any other app you use that supports plugins.
I don't understand what the issue here is or what you want me to learn, those seem to all be bugs that were already fixed. If there are additional unfixed problems that others aren't aware of, and you're not making it up, then please mention those, and maybe I can help you get them reported.
The issue here is that those bugs should never have made it to 'prod'. For absolute die hard fans who say "Gnome or bust", you encounter a bug, you report it, track it, get it fixed, and things are better, until next time the same high level symptom - UI slowness - is introduced, through a different mechanism making it a "different" bug.
Thing is, not everyone is a die hard fan. Developers on Gnome, by definition, are, but outside of that, customers are going to go with what works. And that's not Gnome. Customer's don't care that the UI slowness in version 1 was caused by X, and UI slowness in version 2 was caused by Y. There was UI slowness, which made it unusable, and, well, now they're using XFCE or OS X or Windows. It doesn't matter that there was a bug and it was fixed a year ago. The problem is that there was a bug a year ago, and that particular customer is gone and can't be recaptured - they've got a solution that works and they're not looking to switch to a different one. Especially one that has a history of problems.
This site asks for 1.5 Go of ram https://ubuntugnome.org/, Ubuntu 20.04 asks for 4 Go of ram. Compared to the 30 Mo for IceWM in the article (apple to oranges, I know, but still 3 orders of magnitude), or the 300 Mo for KDE. The author also notes that since XFCE switched to GTK3 they use more resources. That seems to be plenty of basis in reality. Perhaps you mean something else by "resource intensive"?
Keep in mind that the 1.5 GB number is the total amount of RAM you need to have a good time using the operating system, not the amount consumed by the window manager alone. Even OP consumes just about a gigabyte of RAM with all the programs on their Raspberry Pi.
That's not to say GNOME _isn't_ resource intensive -- it is, especially when compared to desktop environments that were either written to be lightweight or written 30 years ago -- that just isn't the best way to measure it.
> Keep in mind that the 1.5 GB number is the total amount of RAM you need to have a good time using the operating system, not the amount consumed by the window manager alone.
That's why I said apples to oranges initially, but you're right, I could have clarified that part.
I find it pretty disappointing that these type of articles and the resulting comments like yours often can't seem to praise something like IceWM or AwesomeWM without also bashing GNOME or XFCE or something else because of "bloat." Who cares? Do you even care? I don't think you do, you'll just delete that desktop and move on and use whatever you want, and that's the whole point.
Also, your characterization of GNOME's rationale is totally wrong. The goal was never to make a desktop with no features, it was more to make a desktop that is streamlined towards certain functionality. If that's not for you, then use a desktop that's more streamlined towards what you want. There's plenty to choose from. What is the real problem?
Edit: I also want to address another common misconception that I see -- the GNOME Foundation does not direct development in a top down fashion and does not impose any rationale on the project. If some feature was removed, it was probably because an individual volunteer working on it decided that it wasn't worth spending time on that anymore, possibly because the need was better filled by a different project.
All I can say is that GNOME 3 seems to take Havoc Pennington's famous essay about as far as it can go, considering configurability a bug. I don't know exactly when it started, but it definitely came to my attention bigtime with the infamous gnome-screensaver.
Yes, I would agree that a lot of GNOME developers still view that as a design goal. But not all do, and even so, having less configurability doesn't mean less features. If you ask me, practically speaking, what it has meant is that there is a larger collection of smaller apps to choose from. So if the app you want lacks a configuration option, you might be better served by trying to find a different app that just works the way you want the first time.
Can you elaborate? Is there something particular that you need that doesn't work on an RPi? If it turns out to be a small fix, then ultimately you may end up caring a lot less than you think.
Also, telling an individual user to "just use another DE" does not solve the problem, whose effects are widespread, on a whole ecosystem of software, its users, and developers.
Could you please explain specifically what one or more of these effects is? I may be able to offer suggestions on how to mitigate it. When I was talking about using something else I was specifically referring to the article, which is talking about using a different window manager, and makes a good case for switching, and may even be able to help you deal with some of those effects. I assume you agree with the article, if not then I don't understand because it seems odd to me to single out this one passing mention in a paragraph that otherwise goes against what was just said, so please elaborate if you can.
I'm not sure what the other user was thinking about, but I think that GNOME's push towards client-side window decoration is annoying and imposing. Depending on your WM, some applications become "useless", they waste space on small screens and just reinforce a sense of fragmentation. I get that it looks better on GNOME, and I appriciate it when using GNOME, but as there seems to be no system to remove the CSD on traditional systems, I still think it is an overall bad move.
Others may find that pushing towards server-side window decoration is annoying and imposing, so this is unfortunately an area where not everyone can be happy. In any case you may be interested to try a patch: https://github.com/lah7/gtk3-classic
Nope. I will continue to bash GNOME because it's something like "the opposing political party." I want Free Software, and by extension, Linux, to gain ground on desktops, and I genuinely believe that GNOME's "philosophy" is not a good one and is detrimental to this cause.
Roughly, I'd characterize it like the following: I believe that the best way for Linux to gain ground is to be a serious, perhaps even a little boring, but definitely "grown-up" desktop. Simplicity is very important, but tt should lean toward configurability and techniness.
What it should NOT do is what GNOME is trying to do, which is trying to streamline and appeal to everyone. This approach is like trying to beat Apple at their own game, which is dumb -- given that beating Microsoft at their own game is well within our reach. Heck, look at Windows 11, we're like a third of the way there already.
I'm afraid I don't understand what you're saying or what the point of your bashing is. GNOME and KDE and XFCE and whatnot are not political parties and are not opposed to each other. They're not really even close to being that, they're open source projects that collaborate on a sizable amount of things. If you believe there is a better way to do something then just contribute to another desktop. The existence of GNOME doesn't prevent you from doing that and isn't detrimental to it, especially when you consider that the cost of switching to another open source desktop is a big fat $0. Please don't unnecessarily politicize things, one of the benefits here is that you explicitly don't have to do that.
To put it another way: maybe some specific mode of "Linux gaining ground on desktops" with "configurability and techniness" is your goal, but people working on any given desktop (including GNOME) may not share that goal in the same way you do, and it doesn't make any sense to expect them to.
Obviously, I can't expect them to. I'm not paying them. But I can absolutely encourage and discourage different ways of thinking and doing, and I'm absolutely free to express that. And yes, I would like to discourage "the GNOME way."
To elaborate a bit about why I'm saying this; For quite some time, onboarding people to Linux was pretty easy. "Just use Ubuntu" was a really good answer. And then Unity came along, with a) bloat and b) this new paradigm that was ostensibly as simple as a Mac, but different enough that people wouldn't recognize it. And this was a terrible direction for Ubuntu to go, especially since XP was on its way out.
And now, when people ask me, hey, how do I get started with Linux, and they've heard of Ubuntu, I can't just say "Yeah, go for it." I have to launch into a thing and name 2 or 3 different distros, primarily due to the weirdness of GNOME. I appreciate freedom and choice, but I'm also quite free to say, I believe the following: I really wish the GNOME people would more-or-less give up, because they have mindshare not off of the quality of their current interface, but because of the momentum of their old one plus Ubuntu. I believe that (much like Windows, frankly) if they actually had to compete in this space on the merits, they'd lose out.
KDE (and LXDE and others) are doing a better job of making a predictable useful interface.
I don't understand what you're trying to discourage, what you're saying doesn't follow. That has nothing to do with "the GNOME way" and is mostly related to the decisions made by Canonical, who decided to make their own desktop, and then dropped it in favor of GNOME. Asking the GNOME people to give up isn't going to make it viable for Ubuntu to develop their own desktop again that is more to your liking. Your issue is with them, not with GNOME. And according to them, you seem to have it exactly backwards: it was Unity that was not able to compete on merits, so it was dropped in favor of GNOME.
But anyway it sounds like you just fixed that by using a different distro, you can even just suggest the KDE or LXDE spins of Ubuntu. Those still exist, and it's not hard to use them. If you're already using them then it's very hard for me to see what your actual complaint is or what you're trying to discourage. It's true you're free to express what you want, but please consider that what you're expressing may not be something that was done with having the full information. I've noticed there is a lot of confusion about what GNOME is and a lot of people seem to mix it up and equate it with Ubuntu or Red Hat or something, which is understandable, but it's not true. I'm only here to explain why.
And just to make it painfully clear: it is already hard enough to maintain open source and fix all the issues and improve the quality of the current interfaces, without people constantly demanding that open source maintainers give up and stop fixing bugs. If you want to get things fixed, you really don't need to do this.
It's not all that complicated. I'm presuming there are people and companies out there working to make these interfaces, to some extent, for other people to use. Maybe it's business, maybe it's love.
Thus, I'm commenting as a fan/user/potential customer, who might find this useful in some way. Or not. Sports fans can talk about their team or the state of the game or what they like and don't like. That's what I'm doing. I'm not "demanding," and if anything, I'm arguing for less work. By the GNOME people, because their thing is a waste of time. Go work on the better things :)
This is incorrect. GNOME users would not consider it a waste of time, just like KDE users would not consider it a waste of time for someone to do work on KDE, or Mac users would not consider it a waste of time to work on macOS, etc. Please consider that your suggestion can be considered rude, you are simply not the target audience.
In general, open source doesn't work like this. These are volunteers, nobody can really tell them to go work on something else, because they work on what they feel like, which may or may not be seeking the approval of people like you. If you want to find another project that seeks your approval, that's great, go do that, let us know about it later and maybe I'll even try it out. If your goal is to harass open source developers until they quit open source, then please stop doing that, that would be making it worse for everyone.
Sooo, I think this is an overly simplistic view, and this is probably the conversation we should all be having, what's driving a LOT of this isn't "what developers choose to spend their time on as something like a hobby." And perhaps it's not the developers who we should be talking about anyway, I'll grant that.
There are "moneyed" business interests who have a lot at stake at their team winning mindshare and profits. Now, I have no problem with businesses being businesses generally. But part of their "product" is these interfaces. And so again, what I would suggest is that pushing GNOME type interfaces is, long term, a bad strategy that is likely to continue the trend of "Linux" being a second class citizen, by wasting energy toward the impossible goal of beating MacOS at their own game.
Conversely, pushing more configurable and slightly more techy KDE-like interfaces I think has stronger likelihood of making "on the tech margin" people more excited and interested in Linux, and ends up helping more people overall.
Now, I could be wrong about this -- but, while I definitely agree with "no one should to demand what open source developers do, especially if they're not being compensated" -- it's equally as bad (if not worse) to suggest "let's not offer big-picture criticism of trends in open source software that affects a lot of people."
Aslo I absolutely accept that what I'm saying is rude. It 100% is.
I adhere to the idea that you should never be rude unless you feel there is something at stake that makes being rude worth it, and I think "trying to gain ground on the Linux Desktop" a thing that affects LOTS of people outside of the developers, is.
> I'm afraid I don't understand what you're saying or what the point of your bashing is. GNOME and KDE and XFCE and whatnot are not political parties and are not opposed to each other.
They surely where born that way, GNOME only exists due to the original Qt license.
Then Gtk+ was always the go to framework for C devs, while KDE was embraced by C++ community on GNU/Linux.
Murray did a very good job for the C++ refugees on GNOME/Gtk, but that has always been a minority versus being on KDE land.
I figured you would post that. I have heard no actual KDE developers saying that's a problem. They don't use Qt LTS. The "drama" there seems to be mostly confined to unrelated internet forums.
As for your posts about C, I can give a personal anecdote: I would agree with those sentiments, C has a lot of really bad problems as a language. But yet, I still write it a lot more often than C++. So It's not as simple as you're making it out to be. Also I checked some of those slides and I wouldn't describe anything there as a "flame war."
>Everything that implies taking sides is politics, no matter what.
If you really want to look at things that way, you could, but this seems to not be strictly true. Just because there is taking sides does not mean there has to be fighting or politics.
Gnome feels like what you’d get if someone read a lot about macOS and wanted to emulate it, but had never actually used it. Gnome makes it very easy for new users to get started and very hard for experienced users to adjust it. Mac makes it easy for new users, but still gives power users room to grow.
Specially in the what concerns the developer experience of macOS, regarding frameworks, IDE and languages.
On GNOME the documentation seems to have stagnated, then we had Anjuta, now GNOME Builder (which tries to be XCode like), a UI design tool that is being re-written from scratch, the various frameworks don't seem to compose, and then each binding does their own stuff, only adapts a couple of Gtk tutorial samples to their own language.
Naturally being GNU/Linux, everything that falls out of the UI only has POSIX or GNU/Linux specific C APIs to refer back to, yet another experience completely devoid of how things work on macOS (there is no Foundation, Accelerate, Core...).
Depends on the year, if you haven't been paying attention they are on their way out, and then we have again the issue of documentation, tooling, language bindings,...
> Gnome feels like what you’d get if someone read a lot about macOS and wanted to emulate it, but had never actually used it.
I find this statement humorous because I've said much the same thing about Linux Desktop GUIs in general. "It's like they were made by someone who's only ever had a GUI described to them in an email". Compared to a system like the original Macintosh or RiscOS that thought hard about how to abstract things for a graphical display and a mouse, or even Windows that just copied as much of that as they could get away with, it's really quite terrible.
Sorry to disappoint you but GNU/Linux is definitely not a third of the way to compete with a Windows 11 desktop, and the best it will ever manage is to run on top of Hyper-V.
> “Who cares? Do you even care? I don't think you do“
> “ The goal was never to make a desktop with no features”
Wow, your first paragraph you accuse people of lying at worst, or capriciously taking a position that doesn’t matter to them at best. It gets better in paragraph number two where you misquote GP and go boxing with that strawman.
So to address your first paragraph, I’d say people here care a lot, and the bashing they do is not out of a place of tribalism, but out of a place of constructive criticism. Most of us here would LOVE to see improvement in $ArchEnemySoftware, and the failures strike deep because this stuff means so much to us. A mixture of disappointment with the status quo, and/or poor judgement calls with design or execution, rather than knocking a bit of tech that doesn’t appeal to us.
As for the second paragraph, GP didn’t say “no features” in reference to GNOME, but decried the basic features that were removed, thereby rendering the software less useful to them.
I’d take a brighter outlook on discussions like these, if the points land on the right ears, perhaps positive change might happen in GNOME. We can hope anyway.
I'm not accusing anyone of that and I'm not misquoting, please stop with this hostility. In my experience, people are happy to switch when it suits them. By "no features" I was referring to those basic features. Sorry for the misunderstanding.
I also didn't see any constructive criticism, if you have any specific suggestions for positive change you'd like to make, then please make them. I'm all ears. But for the record, if you'd like to get involved in development, places like HN and reddit are generally not a good place to do it. The people who are most likely to listen to requests don't visit here.
So let me get this straight. According to you, I should not criticize software that does not meet my needs lest it be interpreted as bashing and ruin the party for those who do use the software in question?
That’s my take, please correct me if I’m not understanding your point properly.
Also, no hostility here. I find these discussions illuminating.
As for Reddit and HN, these places are excellent places for such commentary. Often times the maintainers of different packages heed complaints leveled here and change their ways. That’s a win for everyone if the change is a positive one.
If you have something you want me to understand, please explain. I'm happy to listen to what you have to say. Like everyone else, I'm not perfect, sometimes I don't understand things and I need some help from people like you.
That's not correct, as a show of good faith I've explicitly asked for specific details on what the complaint is. I can't help you if you don't provide those details. If it seems I haven't done that, I apologize, that was not my intention. I'll try to be clearer next time. Let me know if there's anything specific you want me to clear up.
I find it really weird how you seem to think you’re the arbiter of how much people care, when they can share their opinions, and what opinions are valid.
Perhaps you should keep those ideas to yourself and only concern yourself with your own level of contribution to the topic at hand. I don’t need you to mind my comments.
I don't think that at all, I'm just another person giving my opinion, like you. Also, I don't understand why you said that, but then asked me to keep my opinions to myself.
> “ I find it pretty disappointing”
- Then you go into the praise vs bashing.
What do you find disappointing? I guess that’s what I’m driving at.
When somebody knocks MacOS, and it’s a valid point, I hope somebody from Apple is reading! Then maybe it gets fixed. So why would that arrangement disappoint you?
For me personally I find the endless "Linux desktop wars" to be disappointing. It doesn't matter, this stuff is open source, if you really want it to be fixed then just go write a patch. But in my experience most people will just take a few minutes to switch to one of the many alternatives.
I'm just relaying what other people have told me, these type of forums are full of uninformed comments, and are not a good place to solicit feedback. You could hope that maybe an Apple engineer would read it but they probably won't. You would have better luck sending them feedback through their official channels, where you know they will read it. If you're giving feedback to a YC startup then this would be a good place, but otherwise not so much.
“ these type of forums are full of uninformed comments”
Hacker News has solid representation from Big Tech, FOSS projects, and Academia. What data do you have to back up that this is a bad place to air grievances with software?
Whenever anyone mentions Amazon, my ears perk up. If it is within my power to correct, I take action. So what data do you have to support the assertion that HN comments are uninformed, and that this is only a good forum for YC startups?
That's just my personal observation, I haven't seen many unfocused airings of grievances that ended in positive change. Sure there are knowledgeable people here, but they may not always be speaking within their area of expertise; it helps to direct it at someone who is within that area of expertise, in the case of HN, that area is startups.
I've seen plenty of complaints about Amazon here go unanswered. Amazon is a pretty big company, I'm guessing you don't know everything about it and may not be the right person to ask about many things. If I had a question, I'd definitely ask you if it was something you were an expert in, but I don't know what that is because you didn't say.
Thank you for doing that, it's really fortunate that you are, but please understand that not all of your coworkers do the same thing. I think it would be nice if they did, but currently they do not.
Yeah you’re right. But I care about this stuff a lot. I love tech, I love the life it has afforded me, and I want people to have a good experience, even if it is something as simple as billing software!
Obviously Amazon has some huge issues to overcome, both culturally and policy wise. If you keep bringing issues up, I’ll do my best to get them in front of the right folks.
The people that characterize GNOME this way are desperate to identify themselves with a certain sub group of Linux users that prides themselves in doing things the hardest possible way (such as recompiling your terminal every time you want to change the font size) because it makes them feel 1337. Don't sweat it too much, it's just a little bit of garden variety internet tribalism.
I used a Pi 2 as a desktop from 2017 to the end of 2019. Apart from using window maker, I didn't made many changes to use it as a desktop. Also, since it is an under powered machine by today standards, using it as a desktop gave me the feeling of a 90's workstation. I even compiled the latest available GTk 1.x version and xmms for a better retro feel.
It is possible to use the modern web if you keep the number of tabs low and switch heavier sites like gmail to the html mode. An ad blocker is needed and Youtube is watchable if you use the h264ify extension so it uses the hardware decoder and avoids modern video compression codecs.
I could even run libreoffice with Portuguese orthographic and grammatical checkers. It is slow, you can't run other tasks simultaneously but it works.
Want to watch movies? If it is 1080p or smaller encoded with h264 it is totally doable. Even streaming in 720p is doable. You won't be able to easily watch netflix though.
So, yes a 2GB Pi 2 is good enough for light browsing, light text editing, listening to mp3, watching movies and compiling software from early 2000's.
Things you can't comfortably do with it:
- Meetings: encoding and decoding with any codec that is not supported by the hardware is just too slow to be usable.
- Video editing: same as above. Also not enough RAM.
- Compiling large code bases: libreoffice and mozilla will probably fail on anything with less than 8Gb RAM. I could compile OpenCV on it, but it took hours.
- Keep many tabs open: you can if the sites are wikipedia-like. Anything more complex and the system starts swapping. Magic Sysrq-f is your friend in such situations.
- Modern IDE's: no way. Try gnu-nano or emacs/vi. Want to develop GUI apps? Use Lazarus-ide. Anjuta+Glade probably work stably with recent Gtk+3.x versions.
- Games: at the time I used it, the driver didn't support desktop OpenGL. Emulators for old consoles ran great, emulation station ran great, it is possible to run Quake but that is it.
I’ve been trying to use a pi4 as a desktop, with xfce, and was a bit disappointed. It’s usable but less than an old iPad. So I was surprised you got by with a pi2! I’ll have to give that h264ify a go.
For games, I was super pleasantly surprised to rediscover old open source games that I had completely forgot and actually love: frozen bubbles, neverball and monsterz :)
My mindset goes like this: in the early 90's, if I had a Sun SparcStation, I could do all this; a raspi2 is reasonably on par with that, so I should be able to do the same with a raspi2.
What makes it a bit harder is that some software that ran on a SparcStation was carefully tailored to that machine while Chromium is not actually designed specifically for a raspi2... But you can go by with it. This thread gives some hints that I didn't try at the time like changing the cpu governor or using zswap. I didn't even overclocked it out of fear to reduce its lifespan.
But you of course, you're limited by what you can do and how you do it. I even began contributing to gnu-nano because I needed some features that weren't available on it and never got used to emacs/vim, but that is another story.
How do you handle multitasking? Do you shut down what you're doing and then switch to something else? I feel this creates a lot of friction. It could be good in a way b/c you don't get easily distracted :)
I remember in earlier KDE5 version you could suspend activities (in the KDE world this is sorta similar to a desktop - or a group of running applications). This would free up a lot of resources and could mean you can have a lot of applications "open" at once, but they're all in a frozen state. At some point this features was removed. I'm guessing it didn't play nice with a lot of applications. But I haven't been able to find anything equivalent.
You can sorta accomplish something similar by going into the task managed and suspending a task, but it's not ergonomic. Maybe someone has a good alternative?
My multitasking on such machine was limited to a few terminals, Chromium and other utilities. With the terminals, I had one with nano, another ocasionally compiling code and another with mutt; when compiling code, the system slowed down but I could still use nano. Chromium stayed with a few tabs open for querying stack overflow, google or wikipedia and when not using it its tabs were soon swapped out, so switching back to chromium and switching tabs took a few seconds after compiling something. With regard to other utilities, it was mostly hexchat for IRC and xmms for playing mp3. Not much, but I didn't need much more than that. Raspbian is a reasonably quick distro.
For entertainment it was good enough for old games and h264 1080p movies.
You can't really compare the Pi 2 to a 4. That's like comparing a 486 or Pentium to a Ryzen.
The Pi 4 is a serious device. 1GbE, USB 3.0. It demolishes the previous Pi devices across the board in performance. I think this is where people get confused by others using the Pi4 as a NAS or server or whatever. There is this misunderstanding of just how equipped the Pi is today.
> Video editing: same as above. Also not enough RAM.
I wouldn't even do this on my Macbook Pro with 16GB of RAM. Anything less than a workstation with gobs of RAM is just asking for a headache. Unless you're just chopping together the random video for the family once a year.
Unfortunately, the $100 price tag puts the Pi 4 in an awkward place: Not cheap enough to be a hobbyist board, not powerful enough for "serious" things. To me, that's extremely unfortunate, because I think many, many more people need a cheap, underpowered board they can tinker with than need a PC with some exposed GPIOs.
It's $75 for an RPi 4b with 8GB of RAM! And from my experience it's quite powerful enough for a lot of "serious things", including most casual desktop computing needs. And it's only $35 with 2GB of RAM, which is all you need for all sorts of headless server tasks.
It's worth noting there's also some solid competition for Raspberry Pis by other companies making similar hobbyist boards. I'm pretty sure there's one out there that might fit your needs but I forget the name of it.
I heard that things like the Banana Pi (or similar) use horrid chipsets that have very limited support, so they might not actually be of similar utility. If you know of any good ones, though, I'd be grateful!
Nowadays I use an ESP8266 for most things, but it'd be nice to have a small, reliable Linux machine for things like file servers at my parents' house.
> - Keep many tabs open: you can if the sites are wikipedia-like. Anything more complex and the system starts swapping. Magic Sysrq-f is your friend in such situations.
I ran a Raspberry Pi 3 as a desktop for a little while, and I believe I ran into the same problem. In most of my Linux installations, I format the drives without swap. I've never had a problem with it. But, I didn't know how to disable swap on the pi3.
What I would like to see is a tree based handling setup. Everything my bank opens (direct child or descendent): let it through even if it is a usually blacklisted domain because, yiu know, banking is kinda important for me. Same domain via a website I couldn't care less about, blacklist away as per rules.
Indeed - if having such a facility requires me to setup a container per top-level domain (e.g. one container for www.xyzbank.com and everything that it opens), then even that is worth it. Right now, with the combination of ublock origin and other privacy plugins, I don't even try to do financial transactions like banking/flight booking etc for the fear of being left with a semi-processed transaction. That is one of the main reasons of me installing Chrome (I know that Chromium might also be a good alternative for such scenarios).
I guess what I'm looking for is a "allow everything" container setup without having to enable/disable plugins or fiddle with white/blacklists.
What kind of apps that a typical Linux programmer would need to run these days? For me, they are
- terminal (shell, editor)
- web browser (email client, file viewer, etc.)
- clock
With these, I only need a bare minimum window manager (fvwm2 in my case). In my view, KDE/GNOME are made not to scare away beginners and there isn't much productivity gain in using these environments. Once you're used to using Unix/Linux, you can graduate them and start using a more lean setup (while you don't have to, of course). One thing I can see might be handy is the System Settings / Control Panels apps. Other than that (and inertia), what are the reasons that people stick to these environments?
I have the same question. The longer I've used Linux as my main OS, the less I've needed a desktop environment. My setup for the last half-decade has been basically the same as yours, and I've found it to be more productive than KDE/GNOME (or Windows or MacOS) ever were.
I use KDE now, but not much of it really. I do mostly the same things - terminal/browser/some utilities/text editor. I used lighter WMs in the past and they were interesting, but I didn't have time to sufficiently get past the experimental stage with them. Since I really don't do much, it's easier to just get something with easy and stable defaults with only a few changes I need for a setup.
I like to have dvcs and a terminal multiplexer in the VTE. I do email over ssh and pdfs in mupdf which is way lighter than the browser. I also mess with gEDA some.
But yeah, gnome etc is overkill if `sudo vim etc/blah` doesn't scare you and you only speak english.
My gf does freak out editing networks and such and also needs her thai IME to just work so she runs gnome.
I run atom and terminator (a tiling terminal emulator) in my dev environment. Its rather heavy for 'just' development. On the other side it runs without any issues on my T420. Which is a cheap old laptop.
I also love Gnome. For me this is the best, not in your way, desktop environment ever done. For me however :)
Don't you also need to run whatever you are programming? So a compiler/interpreter, REPL, maybe a database, docker, k8s, etc? I assume most devs still want to be able to compile and test locally?
> With these, I only need a bare minimum window manager (fvwm2 in my case). In my view, KDE/GNOME are made not to scare away beginners and there isn't much productivity gain in using these environments
Yeah, everyone who just develops in C and/or web thinks like this. It's one of the primary reasons nobody aside from them uses Linux Desktop.
I've been using lxqt on my Pi 400. It's the closest thing to a "light" KDE that I can find, and I like it a lot. The only problem is that the version on the Raspbian repository is quite out of date. You can get a slightly less old version by enabling backports. I wish there was a good rolling distribution for the Raspberry Pi. openSUSE Tumbleweed is probably the best but it's been glitchy for me and I haven't figured out how to netboot it.
What issues have you had with Arch Linux ARM? I tried it for a little while and things seemed to run fairly smoothly (though many AUR PKGBUILDs didn't indicate support for ARM).
I ran into two problems when trying to switch to a Raspberry Pi 4 8GB as a desktop replacement in mid-2020:
- A noticeably laggy screen update when running urxvt. Any action which caused large amounts of text to scroll on the screen (e.g. a file listing or paging through a man page) would cause a bit of a visual stutter. I was running with with XFT fonts enabled, so maybe that was a contributing factor. Web page scrolling also seemed to lag a bit too if I recall correctly.
- Lack of gamma support for blue light filters such as redshift or f.lux
I'm crossing my fingers there are improvements to these areas in the future.
The general perception of IceWM as a minimalist Windows 95 clone hides the fact that it is indeed a very powerful fully fledged tool for window management. Just icewm-winoptions file and the icesh program enable so much opportunities for automation, it makes the average tiling window manager look tame and primitive.
When I went to senior high back in the early 00s I shared a flat with three friends. None of them had their own computer. I had a spare IBM PC with an old Pentium processor which I lent to one of my room mates. I installed Debian, IceWM and Abiword for her. She had no previous experience with Linux, but with this setup she was able to write her school essays and save them to floppy in order to print them out at school. (We didn’t even have a router in the flat.) It worked really well and was much more stable than Win 98/XP :)
That's not the hard part of using a Pi as a desktop PC...
I've been using them for light to moderate desktop use for years now. The important things in desktopping a Pi, as far as I'm concerned:
- Fix the storage. The SD card is fine for a toy. It will not withstand actual daily use, either in "delivering sane performance" or "handling a lot of random writes without dying in a year." Get a USB to SSD adapter and a cheap 32GB or 64GB SSD from eBay - doesn't have to be fast, doesn't have to be new. It's radically better than an SD card in any use case that resembles desktop use. You can either boot straight from USB, or, to improve compatibility with weird USB to SSD adapters, boot from the SD card, mount root from the SSD, and use the SD card for swap or something (see below).
- Enable zswap if you're not on a 4GB or 8GB Pi4. It should be a module included in the Pi kernel now (I had to argue for a while to get it included). This is not zram - this is zswap. There's a huge difference. It's a compressed swap system that can flush out old or poorly compressible pages to the actual swapfile - which, since you're now fronting it with a compressed swap region in RAM, can be over on the SD card. The bulk of the stuff written to it will never be used again, unless you're just massively overcommitting RAM, at which point nothing really will help. I think around the 5.x kernels, z3fold starts working, though I've not seen a real practical difference between z3fold (up to 3 compressed pages per page of RAM) and zbud (only two compressed pages), as long as same filled pages are enabled - that will crunch a page of 0s down into a "Got it, all 0s, done!" note and not bother wasting space compressing it.
- Set the governor to performance. 'sudo cpufreq-set -g performance' The stock governor is really bad about ramping up, and leads to lag in typing, especially if you're using the atrocities that are modern "desktop" apps (Electron). Eventually, the system will spin up and do something useful, but you improve the performance and responsiveness rather substantially by just pinning the cores to their fastest speed and letting them stay there. I'm sure there's some measurable power difference doing this if you're concerned about the absolute lowest power, but it's not substantial, and "race to idle" solves a lot. If you're using a Pi as a desktop, try it - you'll like the change.
- Use adblockers. It's amazing how much CPU time the internet spends on stuff that is advertising, tracking, and generally evil. Most websites behave a lot better once you remove that garbage.
Otherwise... once you've done that, they're actually quite capable little systems! The Pi4 with 4GB or 8GB is the best option right now, and for most people I'm not sure the 8GB really gains you much - I have one, use it fairly heavily, and rarely see any RAM pressure past about 4GB. But it does make up for some slow disk.
I've written about some of this more extensively over on my blog over the years, if anyone wants more details - link is in my profile.
I should sit down and benchmark those one of these days.
The jump from USB2 to USB3, either with or without UAS (it's nice, but doesn't lead to a huge jump in benchmarks) gets you a ton of disk performance, and my gut feeling is that while you could certainly measure the performance delta from USB3 to NVMe, it wouldn't lead to a practical increase in speed for typical desktop use. Going from glacial SD cards even to USB2 SSDs is a huge jump, but USB3 to NVMe seems like it would be going from "Fast enough that it's no longer a bottleneck" to "Even faster."
Though for reasons I don't understand fully, all the ARM boards I've tested have very poor USB3 performance compared to Intel. Same SSD, same USB adapter, and the Intel systems absolutely crush the ARM systems (Pi4 and Jetson Nano) in small sector IO performance.
The main problem with NVMe is that the various carriers that support it aren't cheap, and often (at least in the past 6-8 months) aren't really available. You can get a USB3 adapter and used SSD for $30, reliably, any time you want.
My dumb, barely related question - how far away is VR/AR from having the necessary pixel resolution to be able to have a fake emacs/vim floating in space three feet from my eyes?
At that point there is a change in the ergonomics and location of coding / most desktop work.
With a Oculus Quest 2 - which has the highest resolution of the affordable VR headsets - it's still a little too grainy for prolonged coding sessions. But I think they'll get there in 1-2 generations.
I'm in the same boat - eagerly awaiting when I can have a VR terminal experience. I chimed in just so you didn't think you were a lone crazy holding out for that.
What is the status of video acceleration on the desktop, wrt surfing the web or playing videos from the browser or a player?
I used the various Raspberries as main media player with Kodi for years with great results, but the desktop experience using Raspbian+XFCE has always been awful, I mean the slugginess due to lack of video acceleration when playing videos, which otherwise is present and works great in Kodi; normal desktop usage was fine.
It should be related to the video acceleration being among the closed blobs that not all distros include, but I thought it had been already reverse engineered before.
My last attempt at it was about 6 months ago though.
Raspberry Pi is a great form factor with popular, documented hardware support for doing weird IO while consuming very little power.
Browsing the web in 2021, on the other hand, requires Watts.
> Using noscript to disable javascript in Firefox has also helped tremendously
It’s this kind of thing that made me switch to a Lenovo M93p for my desktop. It’s a great form factor with perfect Linux hardware support. You can also feel good about recycling someone else’s unwanted hardware.
I really wanted the pi to work out but it just couldn’t handle the JS heavy stuff I use in the day job (webmail, task-tracking, information management / CRM.)
We've been using a Raspberry Pi400 (the computer in a keyboard model) as a media center pc for a little over 6 months now. It plays amazon and netflix videos fine, handles the netflix browsing UI pretty well, but has been rather painful for selecting shows on Amazon for a few months now. Not sure what they changed because it worked great when we first got it.
I'm not sure whether it was the best choice or not, but we figured we'd fall back to a regular pc if we needed to, and so far we haven't been bothered enough to do that.
I feel like as soon as you start using Firefox and Chrome on these highly customised low-powered machines the power/ram saving customisation goes down the drain.
There is something appealing about using a low-resource PC with a limited selection of software installed in order to be free from distractions that could otherwise stop me from being productive and write some code or learn something.
I know the problem is mine from not having discipline and not my modern desktop's fault for being able to run games, but still, I almost want a separate PC for this.
I mostly just used OpenBox for a long time, but my 8GB Pi 4 is running Ubuntu MATE and it works great as a desktop driving a 29” ultrawide (1080p tall).
I’m actually running it from a Class 10 SD Card (external SSDs require unwieldy cases for the place it needs to slot in), and speed is fine for Firefox and VS Code use.
I have a Raspberry Pi 400 and it does a pretty good job as a desktop. I also have a HP x360 laptop from 2017 with Windows 10 and it's HORRIBLY SLOW and constantly kicks on the fan. I'm convinced that cahoots with the hardware manufacturers.
I built a rpi for games and it was OK but never got any use. One thing I ended up using it for was taking to the office and plugging it into my work monitor. With a BT keyboard and phone hotspot for internet.
Ask HN: What do you install this on top of? Do you just start with the vanilla Raspbian installer, the bare bones installer without GUI, or something else?
If you're not going to use the default RpiOS GUI (which honestly is actually very good), you can start with RpiOS Lite and then just add whatever desktop you want.
Love the whole rasp eco-system. Major victory for for FOSS/nix minded people.
I don't think people selling it as a daily driver are doing it a favour though. It isn't meant as such and selling it as such is a disservice to the wider *nix universe.
(and yes I understand that rasp as daily driver in rural africa is literally the projects original goal. that's different though from silicon valley crew attempting to daily drive a rasp though)
It was fun to tinker, but it's questionable if I learnt anything useful other then how to get hobby DEs working on ancient hardware. I guess tangentially I gained an intermediate knowledge of Linux command line tools and some networking concepts.
PuppyLinux [0] ended up being my choice for easy defaults after DSL [1] stagnated. I'm glad to see it's still going and has developed its own ecosystem. Even after I could buy reasonable computers, for a while I took some pride in squeezing performance out of budget computers but ultimately, now that I need to be productive to get paid, it's nice to have computers that "just work" with modern software.
[0] https://puppylinux.com/ [1] https://en.wikipedia.org/wiki/Damn_Small_Linux