Suckless seems like a great way to transition from rice to programming.
A good set of training-wheels for the bicycle of the mind.
However there's something extremely pathological about not wanting well designed and implemented features.
It's like cachexia from voluntarily avoiding healthy fats.
This isn't to advocate feature creep, which is ugly and doesn't add to the design of the system.
In rejecting big systemic components which integrate they may even make things more complex.
This is something like having a small standard library, avoiding RPC, or even loose coupling of components.
Implementing these can reduce duplication, and reduce complexity, while allowing more to be achieved.
Everywhere the minimalist takes something rich and make it small.
>>30075>avoiding RPC.There's also the complexity and performance cost of marshalling and unmarshalling data all over the place which thought recently ran into.
Something like Cap 'n Proto doesn't seem to have this problem.
>>30081>Ironically 8M And Constantly Swapping is pretty light compared to any editor with modern sensibilities.This is true.
>What do you think about the MicroEMACSen and other small clones like qemacs though?These seem to be missing essential primitives, means of composition, and/or means of abstraction. Something like Edwin (a nice Old-English name) is closer in terms of small Emacsen, but alas MIT Scheme lacks the primitives (namely an interface to the outside world) to support a great text editor. (There have been attempts to port Edwin to portable R7RS which with SRFI has a partial interface to the outside world implemented.) Had mg installed on my computer at one point, and it's nice that it was tractable.
>>30092reimplementing parsers for every text-editor in every language doesn't seem like a good idea.
would think compilers would want to make there parsers available for text-editors to use.
another thread full of platitudes and discourse disguised as technical discussion. no concrete criticism or alternative projects, just vibes
>>30105the guys are neonazis and I don't understand why anyone takes them seriously. their stuff is lazy, not minimalist, there is no reason to use dwm instead of writing your own window manager. you can read an xlib/xcb tutorial and write something better (with runtime configuration, for example) in less than a week. literally just go to github.com/explore and search for window managers
open source and "minimalism" appeal to the idea of the lone programmer and inventor, they obviously attract and even foster a petite bourgeois mentality like small businesses. but this mindset is even more aggro and virulent than that of the small owner because the material circumstances of the oss maintainer are more precarious. there is nothing stopping companies or other developers from appropriating their work and making a superior alternative, basically taking over their place. this pathetic position is why they lash out so violently against "the woke", "code of conduct", "trains" and whatever other culture war boogeyman
redundant classification, the culture war itself is fake and a distractionpersonally I claim to be trans on my github and codeberg profiles to keep chuds away from my projects
>>30107>their stuff is lazy, not minimalistYou have to be a programmer yourself to appreciate how cutting complexity at the right places can make software infinitely more predictable and easier to modify. See
http://winestockwebdesign.com/Essays/Lisp_Curse.html for the social problems this creates.
DWMs design could be described as orthodox new jersey, by only coding in the few features everyone needs and writing patches for a couple of others. The status bar is accessible through x11 properties, using dmenu and sxhkd for launcher and hotkey functionality respectively is sound, the lack of any interfaces to window placement is objectively a bad design decision. Ultimately suckless programs are just small utilities done right, they're generally pleasant to work with unless you have legacy requirements, nothing on the level of elegance as s6 or qmail though. My personal window manager is fvwm, it's not pretty but has a nice featureset.
>>30109brother I have written my own window manager, you are just saying words
it is laziness and not minimalism because it lack features and not because it would worsen the code or make it harder to maintain but because they didn't have the energy to write them. another example: the patch thingy instead of an actual module system - it is unjustifiable
>using x11 featuresliterally bad design and it breaks the pretense of minimalism even further. your design ends up conceptually more convoluted because it depends on the arbitrary set of features of a very bloated dependency
Every 2012 minimalism fanatic I knew either went chud or embraced bloat or both.
>>30122trvke
Minimalism is good if it doesn’t go too far. The primary reason why programs are even made in the first place is that they are supposed to be useful. Minimalism is good if it means modularity and lower system resource usage. The st terminal is good example of this because it does what it's supposed to do and it does it well. In theory, it can even avoid some future exploits because terminal escape codes are kind scary.
https://www.youtube.com/watch?v=3T2Al3jdY38 https://www.youtube.com/watch?v=Y4A7KMQEmfo
<flame>It's also known that larger codebase means that's there is higher probability of bugs and it's much harder to even audit or understand large codebases. Just look at Systemd. It does a bunch of unrelated shit, such as…
>sudo (run0)>logs (why not just use existing logger programs?)>acpi (why not just acpid?)>bootloader (why it does have to be a part of a init and process supervisor?)>DNS and resolved? (why it's not standalone?)>and more.Source: systemd documentation and
https://suckless.org/sucks/systemd/Also, it's kinda funny that in Debian, poweroff command is prohibited but you can use systemctl poweroff as a regular user. I want to also point out that the attempted xz-utils backdoor also required a custom downstream patch to OpenSSH that enabled integration with systemd-notify:
https://gist.github.com/thesamesam/223949d5a074ebc3dce9ee78baad9e27If you ask what's the real alternative, then? I would suggest Runit or S6. It's commonly said that those weren't available back when most project adopted Systemd but it's not true because OpenRC and Runit did exists. Some say that Systemd is faster but that isn't actually true because Systemd does something special. The real reason is that sysvinit didn't (usually) parallelize daemon startup but it could be done with sysvinit/daemontools (hack-ish) or Runit or OpenRC (or most other modern init systems). I agree that that sysvinit had to be replaced because it's just not very convenient or user friendly to use.
</flame>>>30197>>30198Tbh I don’t really get the idea of minimalism = low system resources, because since day one of electronic computers existing there has been a desire to provide a virtual machine to act as a more user friendly abstraction to what the actual hardware is doing, even if you’re writing a whole program in modern x86-64 assembly, the instructions aren’t
exactly 1-to-1 representations of what the underlying hardware does and can do with the instructions being represented by microcode and often encapsulating multiple cycles to achieve whatever the instruction does.
Interpreted languages are also as old as compiled languages even despite the extremely tight resources computers had at the time, many domain specific languages rose during the mid-20th century to further provide a simpler and quicker way to get things done on a computer.
Despite a lot of the simping for the old school within computer science and their wizardry for making software with a binary of bytes running on machines with mere kilobytes of memory, it seems like those figures in the field were always trying to make computers as abstracted as technically possible and while the optimisation they achieved in working towards that goal is incredible, focusing on that in the belief that was the minimalism that was sought feels rather like missing the forest for the trees.
>>30197> Just look at Systemd. It does a bunch of unrelated shitSystemd is a deamon suite. No single programm does all of the above.
You may just as well argue that "linux" is a bloated OS because you collect the entire system a single giant unit.
> RunitI use Void Linux for my server and runit made me miss systemd unironically. It's absolutely terrible, but I don''t have the time and energy to migrate to Oracle Linux (I need btrfs).
> S6Tried it once for use in containers. Not a fan of it from what I've seen.
>>30347But a microkernel is more aligned with the UNIX philosophy.
I have heard that the issue with Hurd is that they are perfectionists, not that they can't get a microkernel working. Doesn't harmonyOS (huawei OS) have a microkernel?
>>30220>It's absolutely terribleWhat do you mean by this? The daemontools approach to service supervision is unambiguously superior to cgroups monitoring and other hacks systemd uses (see
http://cr.yp.to/daemontools.html). Runit sits on the lower end of daemontools derivatives, nosh is another one that can run systemd unit files
http://jdebp.uk./Softwares/nosh/worked-example.html Unique IPs: 21