[–]▶ No.957207>>957219 >>957292 >>957298 >>957348 >>957357 >>957364 >>957495 >>957530 >>957547 >>958669 >>959045 >>960176 >>971667 >>972852 >>973104 [Watch Thread][Show All Posts]
There needs to be an operating system built from scratch, and built right from the start. Linux is a big bloated shitball, Windows is a big bloated unstable and closed source shitball, and OS X is like a big cube of gay disappointment. BSDs are all stuck in the 90s and every Unix distro except maybe Solaris is completely dead, and those are stuck in the early 2000s at best. Plan9 is 80s autism.
If you want to know what a really good desktop OS looks like, take a look at Haiku OS. It has a good looking and consistent GUI that just works, and as a result it's simple and convenient to use. The installer works flawlessly in under a minute. Due to it's Unix-like underpinnings, you get a fantastic file system (BeFS) and other tools like Bash, and partial POSIX compliance makes porting a lot easier. It uses a hybrid kernel that's rock solid and offers many of the features of a full microkernel, including the isolation of device drivers.
The ideal server/embedded OS is Sculpt OS. This is a full microkernel system that's based on the seL4 kernel and the Genode framework. This is literally space age software being made available to the public, which was previously only used in the consoles of fighter jets and other advanced weapons systems. You never have to reboot for updates, and it's damn near impossible to knock it offline unless you break the hardware. The modular design allows it to heal itself. If one process crashes, even in the kernel, another can simply restart it without restarting the whole system.
The future is now.
▶ No.957219>>957251
>>957207 (OP)
>There needs to be an operating system built from scratch, and built right from the start.
Said literally every fucking OS developer of all time. Stop trying to reinvent the wheel.
▶ No.957251>>957294 >>957298 >>957302 >>957348 >>957356 >>960742
>has POSIX in any way or form
>built from the ground up
pick one
>>957219
And yet they are all shitty UNIX clones, even Windows. The real next OS will be written in Haskell and all programs in Haskell, so no memory management will be needed.
▶ No.957292>>957301 >>957348 >>972632
>>957207 (OP)
A CP/M clone with a networking stack would be enough for the vast majority of things.
▶ No.957294>>957348
>>957251
>The real next OS will be written in Haskell and all programs in Haskell, so no memory management will be needed.
But then we'll have to rewrite it in rust.
▶ No.957298
>>957207 (OP)
>wanting POSIX
You are a fucking retard, POSIX is one of the main bottlenecks in highly concurrent systems.
The base system doesn't need POSIX or Linux-specific interface APIs, a microkernel would allow for a lightweight container style approach which provides compatibility for existing software.
>>957251
>The real next OS will be written in Haskell and all programs in Haskell, so no memory management will be needed.
And then programs will run like shit, the next big leap in performance for computer hardware will be in the form of high throughput high latency serial memory. Memory management will become a harder for compilers and runtime environments to handle and languages like C++ and Rust will have features added to them to provide the memory layout and management flexibility needed to take advantage of the hardware.
▶ No.957301>>957315 >>957348 >>957382 >>972632
>>957292
Maybe not exactly CP/M, but I like the idea of that. On loonix I can do about 70% of tasks in a framebuffer, and that’s including watching videos, as mpv and some other players allow you to watch in a TTY if given some options. The only issue is that graphical browsers and vidya both require a windowing system. I think it’d be interesting if there was an OS that worked like old DOS, where it was a command line system that loaded whatever GUI application you wanted in fullscreen, and then threw you back to the command line when you closed it.
It would never actually happen, but would be a fun idea. Honestly the easiest and most practical way to go about it would be to find a way to get graphical programs to run on any framebuffer, rather than making a whole new OS
▶ No.957302
>>957251
>real next OS will be written in Haskell and all programs in Haskell
Thanks for the laugh.
▶ No.957309>>957313
>dismissing Plan 9 so quickly
▶ No.957313>>957410
>>957309
>expecting anything good to come out of Ball Labs
▶ No.957315>>957331
>>957301
>On loonix I can do about 70% of tasks in a framebuffer, and that’s including watching videos,
I tried that, but there was no hardware acceleration, and the videos were basically unwatchable. How did you get hardware accel working in the FB?
>I think it’d be interesting if there was an OS that worked like old DOS, where it was a command line system that loaded whatever GUI application you wanted in fullscreen, and then threw you back to the command line when you closed it. It would never actually happen, but would be a fun idea
Be the change you want to see in the world. There's FreeDOS, and you can write programs for that, or you could make your own DOS-like OS with whatever modern features you wanted (multitasking, protected mode 32-bit support, 64-bit support, whatever).
▶ No.957331
>>957315
Use a recent version of VLC you faggot.
▶ No.957348>>957355 >>957356 >>957358 >>957435
>>957207 (OP)
>>957251
>>957292
>>957294
>>957301
>There needs to be
Go ahead, chop chop. Come back when you've written the scheduler.
▶ No.957355
>>957348
This. Everybody knows that the world needs another failed /tech/ project.
▶ No.957356>>957358 >>957425 >>957435 >>957533 >>972859
>>957251
>mandatory stop-the-world gc
>doesn't actually compile to native
Nice scripting language, but Haskell isn't any better than LISP as a real system language.
>>957348
Honestly, we need to realize there's simply too much legacy trash from *N*X/Windows to ever cleanly break from it. There are lots of vastly more modern, superior OSs, but what's truly needed to make them usable an architecture built with some kind of hypervisor environment in mind.
Something like an exokernel OS, with servers capable of running bits & pieces of legacy OSs through shims and APIs, in addition to legacy-free software that didn't need as abstraction.
▶ No.957357>>957358 >>957435
>>957207 (OP)
An operating system moves/edits files, and executes code. That’s it.
Linux is built around the philosophy of “everything is a file”. There’s a lot of fluff but it has a solid foundation. It moves files, and executes code.
What you are arguing is for a more effective streamline of higher level processes, but you can’t beat the simplicity of what we already got.
▶ No.957358>>957363
>>957356
>Haskell isn't any better than LISP as a real system language.
Don't say such things otherwise you will attract him.
>>957357
>t. someone who has only ever written simple userspace programs
>>957348
>Come back when you've written the scheduler.
Thats the easy part and there already exist solutions for it such as seL4, Minix, and what Redox has. The hard part is things like the filesystem, init, a standard lib to allow more complex programs to be written, etc.
▶ No.957360>>957373
I think the hard parts are things like the overall architecture, security, memory management and indeed, some kind of standard library. And then getting things to compile and run if you want to do anything fancy with memory layout or whatever.
▶ No.957363>>957366
>>957358
My biggest gripe with current OSs, aside from the constant need to reboot (New kernel? New drivers? Just change too many settings? Time to reboot!) is their terrible management of shared libs/deps. With memory consumption at multiple gigabytes and rising, this is more important than ever. I want something that can handle multiple versions and forks of multiple libraries, deduplicating things as elegantly, stably, and performantly as possible.
▶ No.957364
>>957207 (OP)
The world needs a lot of objectively good things that would not need any competitors.
▶ No.957366
>>957363
How about splitting libraries up into smaller components, like individual functions and procedures with properly defined specifications and serperate implementations.
Some languages do this and it allows changing the implementation without the need to recompile dependant code, as long as the spec is the same.
That way the only "new versions" would appear when you add new functionality to a library, but these would be added on, and wouldn't impact existing software.
▶ No.957373
>>957360
>I think the hard parts are things like the overall architecture, security, memory management and indeed, some kind of standard library.
With a microkernel approach the hard parts are architecture and the standard library.
Since everything is in userspace the security is inherently stronger since all the individual parts are isolated from one another and its easier to prototype and develop since a bad module can't cause the entire system to die. However since all the parts are seperate processes that have to communicate to each other via shared buffers or through kernel facilitated IPC you need to put more effort into architecting the overall system as any mistakes can be hard to undo.
▶ No.957376>>957377
microkernel: muh minimalism in exchange for virtually any performance. seriously, microkernels use more memory and are much slower than monolithic kernels. have you ever wondered why minimal autists like cat-v haven't remade plan 9's kernel into a microkernel?
because they're fucking SLOW
▶ No.957377>>957380
>>957376
Microkernel based OSs which use shared buffers to communicate between processes are just as fast as monolithic and hybrid architectures.
▶ No.957380>>957381 >>957418
>>957377
Would it make sense do a microkernel together with single address space virtual memory? That way you efficiently share data between processes, since they are all in the same address space. This is of course especially nice for microkernels.
Security would be provided by "proctection domains". But as far as I can tell most research into this stopped in the late 90's for some reason.
▶ No.957381
>>957380
>ZFS for RAM
This is a surprisingly awesome sounding idea
▶ No.957382>>957387
>>957301
Vidya doesn't require windowing at all. Just look at roguelikes, DF etc.
▶ No.957386>>957411 >>957418
>we need the perfect new OS
>let's not support important features because I'm an epic hacker and can do without them
This is why /tech/ will never make anything of value.
▶ No.957410>>957418
>>957313
>expecting anything good to come out of Ball Labs
8 Nobel Prizes for work done at Bell Labs (incl. invention of the first transistors)
3 Turing Awards (At least the first is worth a shit, I'd dismiss the Unix one myself)
whatever you say
▶ No.957411>>957439
>>957386
>we need the perfect new OS
>let's cram tons of useless features that only normalfags would ever need or want
>this will make it better than macroshit wangblows
>amirite guiz
▶ No.957418
>>957410
I was referring specifically to [CY]+3, the Bell Labs of today is a shell of its former self after it was gutted back in the 80s and 90s.
>>957380
>Would it make sense do a microkernel together with single address space virtual memory?
For large scale systems like supercomputers you could in theory unify the address space across all the nodes, however the interconnect between the nodes would be a nightmare since both PCIe and Ethernet/Infiniband fabrics have issues which would cripple the performance such a system.
>But as far as I can tell most research into this stopped in the late 90's for some reason.
Because its fucking difficult and alternate "good enough" solutions came along. I seem to recall reading about a heavily modified linux distro which was centered around the idea but it ran like shit.
>>957386
You can make an OS which does everything poorly or an OS which does a small number of things really well.
Security and stability is something which has to be architected from the start, features can be added to a secure system after the fact. Its kind of like the saying in the video games industry "A delayed game is eventually good, but a rushed game is forever bad" except we are dealing with system security vs feature completeness.
▶ No.957421>>957433 >>957435
>>features only normal fags can go fuck themselves
and tech never made it to mainstream and stayed obscure as fuck idiots
▶ No.957425
>>957356
>Something like an exokernel OS, with servers capable of running bits & pieces of legacy OSs through shims and APIs, in addition to legacy-free software that didn't need as abstraction.
That sounds like something that would need a datacenter to run.
▶ No.957433>>957436 >>958726
>>957421
The mainstream is what ruined everything tech.
Or do you enjoy every website being designed for touchscreens the size of a palm? The CoCs? Maybe it's Intel having a separate backdoor for every intelligence agency on the planet that's doing it for you?
▶ No.957435>>957468 >>957526 >>958625
>>957348
round robbin scheduler
1 line of code
done and less bugs than lunix
>>957356
why do we need to support legacy shit? it's literally all garbage.
>>957356
no, it has a nice optimizing compiler for you to feel good about your best case execution times just like C
>>957357
>files
>solid foundation
>>957421
how is this a problem? literally why would i want normalfags to use my OS? it makes no difference what they do
>muh traction!!!!!111
▶ No.957439>>957467
>>957411
didn't you get banned for avatarfagging?
▶ No.957467
▶ No.957468
>>957435
>round robbin scheduler
Wow, great, something even worse than *N*X's server-focused scheduling.
>it's literally all garbage.
Yeah, but if we don't support it, we have to write and maintain our own browsers, drivers, server stacks, frameworks, programming languages, IDEs, office suites, image editors, video editors, 3D software, games, etc. Unless all you're using a computer for is 1970s-era hobbyist-level hacks you can jury-rig in an afternoon, that's not a realistic proposition. The alternative is today's highly suboptimal experience of VMs w/ PCIe passthrough, or airgap multiboxing.
>it has a nice optimizing interpreted runtime emulator
Yah nah
>literally why would i want normalfags to use my OS?
To subsidize development and perhaps break a few out of their stupor
▶ No.957490
▶ No.957495
>>957207 (OP)
That's why Linux from scratch exists you stupid nigger.
▶ No.957526
>>957435
>why do we need to support legacy shit? it's literally all garbage.
It's tried and tested garbage vs your not tried and untested garbage.
The world is full of retards claiming to be the smartest man in the world, and without legacy support you would need a lot of effort not to look like one of those.
▶ No.957530
>>957207 (OP)
>There needs to be an operating system built from scratch, and built right from the start.
Yea but if you didn't write the OS then you don't really know what's in it.
And if you didn't build the chip then you don't really know whats in it.
These are the only two problems with your plan.
An alternative would be to build a 8-bit computer using an Arduino or similar processor.
It would be slow and you would need to write all the software yourself, but it should be capable of simple web/email use.
And at least you know it wouldn't be susceptible to any common attack or data leak.
▶ No.957533
>>957356
>doesn't actually compile to native
It does. But it's still an absolutely horrible systems language. Maybe a future language in the same family will be good enough one day, if type research can obviate exceptions and solve the problems with GC and performance.
▶ No.957537>>957543 >>972904 >>973063 >>973128
You people talk shit and act as if you know what's good and what's bad.
Everytime I come to this place I am amazed by the level of ignorance that you exude. None of you are willing to STFU, sit down and make your own CPU, motherboard, compiler & OS. It's not impossible. I did it, and so can you.
Patience, sharp skills and knowledge - they will be polished during development, testing, and debug - are required and it is a LOT of fun. And, RTFM the standards to write your own drivers, apart from networking and USB the rest is easy.
You can make a 1000 x 1GHz Dual core chip - my own CPU RTL - in China for about 25.000$ + 10.000$ for the 1000 x motherboards (PCB & parts) - my own circuit design.
Beware of shipping quantities, you gotta split the packages to avoid customs if you're not going to sell. The Chinese have no problem with such arrangements.
You can also use an FPGA or an array of FPGAs on which you can implement your own CPU architecture and use other FPGAs for I/O and ...
As a matter of fact, my router is my own design implemented on an FPGA and I'm posting from my own board running my own browser.
Intel, ARM, AMD, ... are flopping big time. We ALL know who they serve.
Stop this shit-talking non-sense and get to work. Otherwise, STFU!!!
▶ No.957543>>957549
>>957537
Why don't you sell us your CPUs, Terry?
▶ No.957547
>>957207 (OP)
>monolithic kernel - monolithic = bad
>microkernel - micro = good
Y'know if it was Microkernel vs Megakernel you fags would think megakernels were better because they have the cooler name. This is literally the thinking ability of computer nerds.
▶ No.957549>>958630
>>957543
This is no joke. I'm not Terry but I wish I could spend some time with the guy, he's unstable but not crazy.
Because I'm not Yehuda (nothing against Jews), I won't sell anybody's soul for 30 gold pieces. I did what I did for fun (also a PoC - I live alone, have no friends and not ineterested in people's egos - a VERY private person) not for profit.
What is money compared to your soul?
Eventually, when I figure out how not to get fucked by some patent holder or a troll, I'll publish everything online. Do it properly or don't do it at all!
▶ No.958593>>959044
Sadly people's demands for computers is so numerous and evolving so fast that no one could possibly start from scratch and actually get anywhere in terms of mass (or even minority) adoption. Imagine making an OS knowing that you have to be able to perform all the countless tasks expected of a modern computer due to modern tastes.
▶ No.958625
>>957435
IMO, best bet is build something bottom up, correctly, and let any legacy be rebuilt to suite.
How many times you do backup your files, install a new os, and never touch the backup files again?
▶ No.958630
>>957549
Put at least a dead man switch to release everything if you do have an early departure.
PSA:Patent troll always find a way to fuck you.
▶ No.958658>>958797 >>958823 >>958897 >>959044 >>960743
I've been interested in osdev, the problem isn't programming the kernel, its designing an ecosystem and challenging the current paradigms.
without strong insight you end up with a variant of unix, dos, or vms/nt.
it would be nice if there was a bare bones kernel with some memory management, scheduling, and paravirtualized drivers as a starting point for people to experiment with actual design.
I'd like to see a modular gui system, were applications are built like shell scripts in a visual block diagram out of small simple programs, break the tradition of plain text ipc, have simple concurrent models, and have a language tower like racket were programs written in different languages have a shared interface (no need for language bindings, every tool is written in the appropriate language).
my os would be focused on making computing fun rather than practical.
▶ No.958669>>958797 >>958798 >>958904
>>957207 (OP)
All I want is to boot my computer and get BASIC (not boot linux/win/whatever and then launch BASIC, just plain old interpreter doing all the job). Write your own source code, share it around, use it for what you need to do and that's it. Special points if it's able to interpret multiple programs at once.
I'd go for that.
▶ No.958692>>959044
>>957387
You don't need the OS to have window system to make those kind of games. All you need is a way to draw on the screen.
▶ No.958720>>960238 >>960751
I'm sure our hero would say this.
▶ No.958726>>958817 >>958999
>>957433
Microsoft does that as a courtesy, not because NSA even needs them to. PRISM ensures that NOTHING YOU TYPE IS EVER GOING TO BE ANONYMOUS.
Ironically, the people that are the biggest threat to society (organized crime) are the most immune to PRISM, via gigantic botnets that ENTIRELY hide their activities. Literal needle in a hay stack. Metadata doesn't mean anything if 50 million innocent people are tagged incorrectly via botnet.
This is the current state of 'the war on cybercrime'. Futile. And any time some one smart enough to track those fuckers comes along? They're fired and hunted, because it means the NSA will lose funding for their bloated, inefficient bullshit.
Spying on you is more important than actually stopping crime. The CIA did the Las Vegas shooting.
▶ No.958797>>958798 >>972681 >>972763
>>958658
>a bare bones kernel with some memory management, scheduling, and paravirtualized drivers as a starting point for people to experiment with actual design
>>958669
>All I want is to boot my computer and get BASIC (not boot linux/win/whatever and then launch BASIC, just plain old interpreter doing all the job). Write your own source code, share it around, use it for what you need to do and that's it. Special points if it's able to interpret multiple programs at once.
why doesn't this stuff exist? i'm always going down this old computer rabbit hole. i think about getting an old, simple computer. like a commodore 64 or spectrum. very quickly it becomes problematic because: 1) old hardware may be more prone to failure, 2) you would need floppies that they no longer manufacture.
it's such a shame. those old computers had some problems, but they got a lot right. maybe the next best thing would be a bare bones os.
▶ No.958798>>958804 >>958921
>>958797
>>958669
Install Linux and a command line interface a text editor and a compiler. You can do this today so what's stopping you?
▶ No.958804>>958806 >>958811
>>958798
first of all, i don't like linux. i use openbsd and use text based programs.
that is still not a substitute for a simple system like the commodore. i should not have to explain why. if you think the appeal of a simple system has to do with simply using text programs then you are misled or confused.
▶ No.958806>>958810
>>958804
Operating systems exist in the first place because most computer users like the multiplexing capability of a computer and also for hardware driver functionality being available for all software. If you want to go back to the age where one computer can only handle one program at a time where each program is required to independently write their own drivers, that's certainly possible but that's a very strange way for general purpose computers.
▶ No.958810>>958815
>>958806
why is it all or nothing with some people? you can often have both.
just because i would like a ford model t to drive around on a sunday doesn't mean i don't want a modern car to drive to work on a winter morning.
also, some of the things you have said make me think that you don't know a whole lot about computers.
▶ No.958811
>>958804
Start reading the developer guides for your architechture of choice. Learn some assembly. I'm assuming you know C. It shouldn't be too terribly hard to write a little basic interpreter kernel, especially if you only want it to be single threaded.
I looked into writing my own microkernel awhile back. I read through the dinosaur book. It's good for learning how to call the kernel and do multithreaded programming, but I should have read Tanenbaum's OS design and implementation. It's on the to do list. Also you can check out the OSdev.org wiki.
▶ No.958815
>>958810
You're telling me that you want a simple system and you define this as a Commodore 64. The Commodore 64 didn't have an operating system because nobody bothered to write one for it. I'm saying you can go back to that simply by doing the same thing - write your program directly for your computer without an operating system in it. Doing your computer this way is so trivial that any person trained to write hardware drivers is capable of achieving it in a short time (a couple of weeks). I'm confused about why you bitch about how software is too complex when you could be implementing this system by yourself.
▶ No.958817
>>958726
>says a bunch of harmless things because literally not a single person with a brain would be unaware of them
>they are still true nevertheless
>throws one wild preposterous theory as fact at the end so everyone can see he's insane
>job done, well of relevant info is poisoned, discrediting valid facts operation complete
▶ No.958823
>>958658
>it would be nice if there was a bare bones kernel with some memory management, scheduling, and paravirtualized drivers as a starting point for people to experiment with actual design.
On x86 you don't have to write more than 1000 lines of mostly boilerplate for that as the hardware does almost all the work if you do one task per TSS. I don't know if x86_64 requires more from the programmer but it's not like you need a 64 bit kernel just to experiment with. I assume there are still tons of resources on getting started as we all used to write our own protected mode loaders (essentially, mini operating systems) back in the DOS days before dos4gw got popular because we had no choice.
The hard part used to be that there are multiple things that have to go right on switching to protected mode and if you fuck one up there's (almost) no way to communicate that to the programmer as ram is in an unusable state until that switch is performed correctly so you're left guessing. I used to flash the keyboard LEDs to see how far my code got before the dreaded triple fault as they didn't require access to memory. Today you can just attach a debugger to qemu and spoil all that fun. Use "legacy BIOS" or whatever they call it rather than UFIA UEFI mode so you don't have to fuck around with bringing up controllers and all that modern bloat just to write to the disk or put some text on the screen.
▶ No.958897
>>958658
RISC-OS has a pico image that boots right into BBC BASIC.
▶ No.958904>>958921
>>958669
Google Colour Maximite
▶ No.958921
>>958798
>Install Linux and a command line interface a text editor and a compiler.
Or FreeDos, or whatever. But:
>not boot linux/win/whatever and then launch BASIC, just plain old interpreter doing all the job
The fewer layers, the fewer opportunities to dig hole in security. I don't want an overly bloated product that takes hours to install, just a program dedicated to a specific hardware and entirely programmable.
I guess it would have to be able to both send ASM instructions and interpret its own language. Possibly handle two types of libraries: one for language keywords (written in ASM), and another code library for applications (written in the same language as the source).
>>958904
Nice! Thanks a bunch.
▶ No.958926>>960235
imo you'd rather LARP than write code.
▶ No.958999
>>958726
>PRISM ensures that NOTHING YOU TYPE IS EVER GOING TO BE ANONYMOUS.
We'll fucking see about that motherfucker.
▶ No.959044
>>958593
>>958658
This is the motivation behind my pet concept of a modern OS built around hypervisors/APIs. So that it would both offer a superior user experience for legacy bullshit out of the gate, and the prospect of future "native" software offering an even better experience.
>>958692
The benefit, though, is your game coexists better with other software, instead of being a giant (usually mono-window, games a shit at multi-monitor) blob.
▶ No.959045
>>957207 (OP)
>There needs to be X built from scratch
Lemme guess. You offer to contribute a logo? Faggot.
kys
sage
▶ No.960164>>960236 >>960357 >>960713
I'll get started on the logo :^)
▶ No.960176>>960229 >>960408
>>957207 (OP)
Or you could use an actually successful microkernel called QNX. It works on phones, embedded devices, and desktops.
▶ No.960229>>960231
>>960176
>just use this propriety microkernel as the base for your new OS!
no thanks, also your implication that seL4 isn't successful is funny
▶ No.960231
>>960229
>your implication that seL4 isn't successful
it's not
>use this propriety microkernel
it's literally the most reliable OS that's not just a hypervisor like all the other meme "microkernels".
▶ No.960235
▶ No.960236
>>960164
Don't forget the mascot!
▶ No.960238
>>958720
le pragmatism XDD
▶ No.960357
>>960164
I'll draft up the code of conduct.
▶ No.960408
>>960176
There's lots of rock solid little RTOSs along those lines, like OS-9 and VxWorks.
▶ No.960713
>>960164
I'll draw the mascot :P
▶ No.960742
>>957251
I was rooting for Lindows and OS2warp, early on. I stumbled on linux just because it was free
▶ No.960743
>>958658
Thoughts on TempleOS? I've been debating learning osdev off it and try to get some sort of virturalization into it.
▶ No.960751
▶ No.960816
How about you start with a processor that is not backdoored
https://www.youtube.com/watch?v=GQE3a8WLeds
▶ No.960832>>972632 >>972919 >>975646
>>960818
When will Google release Fuchsia already so the market share of the Linux kernel can sink like an anchor and freetards can finally kill themselves?
▶ No.971658
▶ No.971667
>>957207 (OP)
>Linux is a big bloated shitball
Kill yourself bet you never configured your own kernel
▶ No.972616>>972635
▶ No.972632>>972675 >>975646
>>957292
>CP/M clone with a networking
Isn't that just DOS with an ethernet card installed?
>>957301
You don't really need a windowing system for either of those. You need it for multitasking, but not the programs themselves.
>>960832
>thinking google is any less pozzed
They're done far worse than cave into the CoC bucko.
▶ No.972635
>>972616
Yeah, I've been looking into SculptOS that OP mentioned. So far I like what I'm seeing. I just wish seL4 had some Spectre/meltdown mitigation. I guess it's time to hit the books.
▶ No.972675>>972680
>>972632
I think when he says networking he means TCP/IP, which DOS didn't have.
▶ No.972680
>>972675
Certainly not built in, but nearly every DOS ethernet driver contains some implementation of the TCP/IP stack or at least IPX. At least from what I've seen. Not to mention the numerous third-party IP stacks available.
▶ No.972681>>972682 >>972690
>>958797
> you would need floppies that they no longer manufacture.
Just about every major "retro" computer platform has a floppy emulator project that supports it.
▶ No.972690
>>972681
I'm really hoping it becomes feasible for hobbyists to start up manufacturing of those old discs and ICs though. Having that all lost to failures over the decades isn't something I'd like to see. Archiving ROMS/disc images and emulators can only do so much.
▶ No.972714
I say write the next os in brainfuck, that would push the limp wristed faggots and "femanasties" away do to the vulgar language.
▶ No.972763
>>958797
Maintaining the hardware is part of the fun, and in reality is simple and rarely needs doing. The worst that could happen is you have to replace some ICs that are no longer manufactured, but usually you won't have to do anything more complicated than replacing filter capacitors.
Also, as others have mentioned floppy emulators are common and available for most if not all platforms.
If you're still worried you can still build one of dozens of new projects based on older ICs that are still in production.
https://retrobrewcomputers.org/doku.php
http://www.malinov.com/Home/sergeys-projects/xi-8088
http://mtmscientific.com/pc-retro.html
▶ No.972764>>972877 >>975646
>>960818
>fuchsia
>fucks ya
How obvious can it get. This is going to be the next step in their master plan.
Everybody is already turned into a mindless robot by android. Now they are going to finish us all of.
Choosing such a name is typical jewish tricks. Same as the various 9/11 foreshadowing.
▶ No.972852
>>957207 (OP)
It's what I tend to say. I want to learn how it would actually be done, but I don't have the knowledge that learning would require in the first place, so I'm working on that first. I have tried a lot of software, and nothing really satisfies me.
▶ No.972859>>972892 >>972906 >>973600
>>957356
>Nice scripting language, but Haskell isn't any better than LISP as a real system language.
Shitting on Haskell as a choice of OS language is one thing, but Lisp Machines were some quality computers back in the day.
▶ No.972877>>972902
>>972764
It's pronounced few-sha, idiot.
▶ No.972885
Doesn't Haiku use FreeBSD code? Is FreeBSD's pozzing not a concern?
▶ No.972892
>>972859
Such quality that even shit tier cheap x86 chips were faster.
▶ No.972902>>972913
▶ No.972904>>972908
>>957537
>1GHz DUAL CORE CHIP FOR 25.000$?
nigger what the fuck are you smoking holy shit, hahahahahahah
▶ No.972906
>>972859
Lisp machines were always garbage. They were mostly owned by universities and there were only a few thousand ever made.
▶ No.972908>>972911
>>972904
fabrication cost you retard
▶ No.972911>>972917 >>972928
>>972908
I know faggot, that cost is just so far under market value its retarded
▶ No.972913
>>972902
Not in English. It's pronounced "fyoo-shah."
▶ No.972917>>972918
>>972911
I dunno. You can get a raspberry pi for 35 bucks, and that's got a 64 bit quad core ARM with wifi, bluetooth, ethernet, and 4 usb ports.
▶ No.972918
>>972917
Tooling costs for a piece of silicon at that complexity is several millions. A dual core CPU thats running at 1GHz would be at the very least 500k in tooling and upstart costs, but don't forget you need testing, packaging and if you fuck anything up you have to pay for a new design. To think anyone other than a super genius, mega autist millionaire could do this is insane.
▶ No.972919
>>960832
Why do you hate freedom, Pajeet?
▶ No.972928>>972936
>>972911
Nah, should be doable. Single core 64 bits 1ghz arm chips can be had for way less than that, like 17 dollars a piece.
▶ No.972936>>972970 >>972974
>>972928
Yeah, 17 bucks when they manufacture and sell millions of them. You obviously don't know anything about silicon manufacturing if you think market price for a part is in any way connected to manufacturing price, unless its some niche device or something with a huge silicon area.
▶ No.972970>>972989 >>973117
>>972936
Hey, what about bootleg chips? We are talking about the Chinese after all.
▶ No.972974>>972989
>>972936
It is obviously connected in the sense that manufacturing cost is below sales price.
Masks are a huge cost, but it might be significantly less for an older process.
And cost can be split among multiple customers, otherwise universities would not be able to afford making prototypes.
▶ No.972989
>>972970
Maybe 200k for the type of chip he is talking about. A dual core CPU on old process technology is going to take up some mm^2, and the chip also needs to work so a shitty chinese fab with high fault rate on manufacturing will be an extra expense.
>>972974
He said it was 25k for 1000 chips, that doesn't seem very compatible with shared wafer solutions. That price is just completely unbelievable to me.
▶ No.973063
>>957537
Did.. did you solder it and shit?
▶ No.973104
>>957207 (OP)
I'll make the logo.
▶ No.973117
>>972970
Most processors are fabricated in Taiwan, not China. They are two different countries despite what China wants you to believe
▶ No.973128>>973134
>>957537
>Intel, ARM, AMD, ... are flopping big time. We ALL know who they serve.
Most Intel chips are fabricated in the USA and Costa Rica, most AMD chips are fabricated in the USA albeit GlobalFoundaries is now owned by the Saudis and most ARM chips are fabbed in Taiwan and ARM itself is owned by the Japanese
You people love conjecture and spreading disinformation for some reason
▶ No.973134>>973152
>>973128
Pretty sure the majority of chip foundries are located in east asia or isreal these days.
▶ No.973152>>973155 >>973600
>>973134
You are incorrect please stop doubling down and read a book.
▶ No.973155>>973600
>>973152
>You are incorrect please stop doubling down and read a book.
Oh, so you're larping as /leftypol/?
Have fun then.
▶ No.973600
>>972859
>hardware-accelerated stop-the-world gc
LISP machines were ridiculous on an architectural level
>>973152
Over 70% is in East Asia:
http://anysilicon.com/taiwan-maintains-largest-share-semiconductor-wafer-fab-capacity/
>>973155
>ignorantandproud.bmp
▶ No.975646
>>972632
>>960832
ho ho ho
https://github.com/fuchsia-mirror/zircon
https://github.com/fuchsia-mirror
>>972764
The secret OS meant to replace Android has a big problem to overcome: Google itself
https://bgr.com/2018/07/19/fuchsia-vs-android-gogole-os-development-details-leaked/
>The company must also settle some internal feuds. Some of the principles that Fuchsia creators are pursuing have already run up against Google’s business model. Google’s ads business relies on an ability to target users based on their location and activity, and Fuchsia’s nascent privacy features would, if implemented, hamstring this important business. There’s already been at least one clash between advertising and engineering over security and privacy features of the fledgling operating system, according to a person familiar with the matter. The ad team prevailed, this person said.