“If we were to re-write Linux, taking clues from various operating systems, what would we make sure it did? Our next-generation operating system (NGOS) would be completely modular in design, aimed at 64-bit hardware, and with an interface that would change the way people compute. It would support a large number of applications and hardware devices, accepting device drivers written for other operating system, and run applications written for other operating system under an emulation mode.” Read the editorial at NewsForge.
If I Could Re-write Linux
About The Author
Eugenia Loli
Ex-programmer, ex-editor in chief at OSNews.com, now a visual artist/filmmaker.
Follow me on Twitter @EugeniaLoli
The Kernel and GUI points are the best.
Personally, I just want an installer framework for Linux like Mac OS X or Windows XP.
Many of the points are just plain wrong, and in any case it’s pointless castle-in-the-air, not real design goals or possibilities.
I would had an HAL to the kernel.
If I in the future create a operating system won’t be GPL but a commercial operating system, because it takes me years to develop one, and I would want to make profit from it.
I thought OS/2, BeOS etc. tried that. And look where they are right now 🙂
Btw, the guy that wrote the article is moron. Technical aspects have nothing to do with the sucess rate of an OS.
It’s what you do with it after it’s done is what matters.
Looking at the Windows family, I think there are a few key apps and technologies that ensured it’s success: Access, Excell, VB, Word, ASP, Games
I would follow Jef Raskins ideas;
No files. No applications. No modality. No loss of data.
Just a modeless, fully integrated, searchable, zoomable world of commands and transformers.
If you have no idea what I am talking about, read his book, or start with his website (http://www.jefraskin.com)
That should be the title of that article. Otherwise, it would take another 10 years and by that time most of ideas will be not so modern anymore, like 64bit OS or file system. MS is dreaming about db like filesystem. Who knows what will happen in 10 years.
I was giving this article serious consideration until
I read the following,
“NGOS would also follow a modular design but would
do away with the shell prompt altogher”
Why get rid of the shell prompt? As I read Unix history,
the shell prompt has been convincinly demonstrated as the
most flexible and powerful way to control the OS. Provide
alternative, sure, but get rid of it? This indicates to
me that the author has a superficial acquaintance with Unix.
Made me think of the quarrel between Linus and Tannenbaum
http://www.dina.dk/~abraham/Linus_vs_Tanenbaum.html.
Otherwise I agree. Nothing much new. But the author invites to tag on to the list. Files and other objects in an SQL like db sounds good to me.
The author of this article definately makes the mistake that “Linux” needs to be as like Windows as possible in order to be useful.
Useful to whom? The proposed changes would only make my system less useful to me.
Progress is a tough word to define, but I see it as being entirely dependent upon a projects goals. For Linux, progress is increased hardware support, more platforms, etc. For Mandrake, progress is more automated hardware support, easier package maintenance, and easier configuration. For Gnome, progress is better usability, and more sensible, minimalistic configuration. For KDE, more features is the goal.
Please, don’t lump an entire software platform under the “trying to beat Windows” umbrella and thereby impose design decisions on me that I may (and do) entirely disagree with.
Don’t even waste time reading this. It’s like he’s copying and pasting people’s comments from listservs (of mostly ex-BeOS users). Seriously, writing an OS is not nearly so cut and dry, you should try it.
But, let’s say, if you actually could rewrite Linux from scratch, most of the specific details from the article would be obsolete by the time you finished. I mean, let’s face it, Linux is at least 12 years old. In 12 years, we won’t be using 64-bit machines anymore, we’d probably be using 128 or 256-bit, or heck, maybe even organics or trinary or fuzzy state machines.
By then, who knows, the game of Go could be “solved” (or some other equally difficult problem), ushering in a new era of problem solving where we devise models much closer to the way men and women think, forcing software to mould and shape itself to a whole new set of constraints.
Sarcasm aside, it’s like throwing out the water before it ever has a chance to boil. It’s much healthier to aide already founded projects like Linux or Gnome, unless they simply -do not work-. Just because the Linux kernel is monolithic, for instance, doesn’t hurt it. It still works, and it’s still worth putting time into to make better and better. For the counter example, look at how the 2.6 kernel can be scaled down for smaller architectures. That’s what you want.
Fix what doesn’t work, don’t throw it away.
“I would follow Jef Raskins ideas;
No files. No applications. No modality. No loss of data.
Just a modeless, fully integrated, searchable, zoomable world of commands and transformers.”
I think Raskin’s ideas are suitable for a modest PDA or a very small home computer like the original Mac.
They won’t work on a computer for serious CAD, programming, music composition, publishing, games design, etc.
Behold, the clueless reporter…see him in his native environment.
“Even though Linux has a monolithic kernel it has still been successfully ported to more platforms than any other operating systems.”
Bzzzt. Wrong. NetBSD has :]
“NGOS would also follow a modular design but would do away with the shell prompt altogether. Its GUI would talk directly to the operating system.”
Exactly as XFree does. It talks directly to the kernel.
“Users who still like a command prompt could execute a command shell such as bash and DOS on top of the GUI.”
Uhm. What do you think Xterm, Eterm, gnome-terminal, konsole, etc. do? Twiddle their thumbs?
This man knows extremely little about operating system design.
Do yourself a favor, skip this worthless drivel.
“I think Raskin’s ideas are suitable for a modest PDA or a very small home computer like the original Mac.
They won’t work on a computer for serious CAD, programming, music composition, publishing, games design, etc.”
Have you read The Humane Environment??
I’m surprised you can say that if you have. I really think his ideas are scalable to the amount of stuff we do with computers today, and I see it fitting much better as my desktop pc than my pda.
I don’t see a single thing he says preventing me from doing any of those tasks you mention.
So please stop posting these ridicluous ‘if only’ pieces, and start doing something about your perceived problems.
So clueless it made me laugh: we need a 128bit filesystems of course 64bit won’t be enough 🙂
I don’t think that the author understand how big 2 power 64 is..
I think a lot of people are missing the point of this article. I think it is ment to stimulate conversation about future features that OSes should incorporate, or at least strive for.
I agree with him about the shell prompt. Get rid of it and boot directly into the gui. If you want a command line, fine set it as your shell, I do agree that a command line can be easier in certain cases, but guis are much more intuative, as long as it is labled logically. There is a reason that Apple decided to ditch the prompt , and I think a lot of people are forgetting that.
Be got most of it right, but then ran out of funding.
People who think that “Linux” can be directed at any ends are missing a bigger point–all sorts of different people use free software to all sorts of different ends, not just what’s easy for mainstream PC users. There are many other markets to cater to. Desktop GNU/Linux distributions will adapt to however it’s users find themselves computing.
Who knows what the NGOS will look like? We don’t even know what computers will look like.
This article is a good joke. To sad, the author never used a computer, I guessed it is a joke too fast.
BTW, I do OS design for embedded devices. If you want to do solution for end user, take lessons about cognition first not computer science. GUI is about dealing with people not computers.
See you in moderated down comments.
Why would you want to load a ton of resources for no reason at all. A GUI robs the system of resources. I like having the “option” of a GUI, but not the requirement. Why not just run Windows?
Why re-write Linux? How about re-writing MacOSX? Or Microsoft Windows? Each OS has their problems. May be Linux could have a better user friendly side. May be MacOSX could be run on more hardware. Maybe Microsoft Windows could be a bit more secure.
The article should have been written a different way. Title the article as “If I Could Create My Own OS”. That way it seems less like flamebait.
I am surprised that people use Linux considering they want to re-write it, change just about everything that makes it work, and alienate the people who started it. Unix does have a tradition much like MacOS and Windows. If you want to get a feel for the Unix history read The Art of Unix Programming by Eric Steven Raymond. It is free to read online or you can by the book.
Here is a link:
http://catb.org/~esr/writings/taoup/html/index.html
has most of the charecteristic in the teaser.
I’ll license it via BSD or GPL, a _good_ programmer can make money even if his works are BSD/GPL licensed.
Personally, I just want an installer framework for Linux like Mac OS X or Windows XP
Yep, that would be a major step to get linux into the mainstream. That is the main thing holding it back now. The RPM system is ok, but a installer like windows xp would be nice. It needs to be click a .rpm, it would ask you the root pass if you are not already root. but then it would ask which users to be allowed to use it, then actuelly put a shortcut on the desktop &/or menu. Well it should ask about conponents to install if it has more to install also.
I red da artical and it so stupid I tink we all got a little bit stupider after reding it.
The author seems to be missing something large here… He wants to take what is arguably a great server system and make it into a great desktop. It doesn’t work that way. Microkernel and 64 bit are obvious steps in evolution, and everybody loves more storage space. However, the file system he describes (whereby, you rename folders “containers”) and the lobotomized interface are far more befitting an end-user system than a server OS. Command line is a very convienient way to script a production system, and should be expanded to be more like a programming language, rather than removed altogether.
I like the idea of clean, unfettered desktop OSes. I like the idea of focused, workhorse server OSes. I don’t like the idea of a workhorsey end-userey OS. At best you get the desktop of Windows with the power of Linux. At worst you get the desktop of Linux with the power of Windows. Either way, if you are trying to replace one or the other, you should focus on that target and not trying to be all things to all people. For example, X11 is a powerful remote windowing system that makes Linux a viable solution for large computing center applications, but for the desktop it is a dog.
You can have a great server OS, you can have a great desktop OS. You can have a passable server desktop OS. Choose one.
I agree with him about the shell prompt. Get rid of it and boot directly into the gui.
Well, most Linux distros are doing that right now, no?
I’ve always had a big problem with Linux:
it’s an old generation operating system (i.e. it is really just an incremental evolution of the unix architecture, it is not a forward looking operating system), yet it has so much effort dedicated to it, and people seem to be forgetting about looking at a real next generation architecture. I could have told you this 10 years ago in my operating systems lectures – and yes, I was using Linux 0.95.
The single most cumbersome thing in a windowing interface is having to scroll something vertically *and* horizontally to see it. There’s no smoothe, logical way this can be done. You guys with the most high resolution monitors might not see my point, but there are plenty of webpages that are always two inches too wide for the screen. I don’t care if it’s an issue of screen drivers or whatever, I don’t care, this left-to-right scrolling is insane. Get an old copy of GEM 2.0 and see how different navigating a directory is when left-to-right scrolling is not even a possibility, because it’s natural, just like with an actual scroll, your eyes go left to right, but your hands only scroll the page vertically. Very natural, that’s why scrolls were made that way.
could there be any more buzzwords in there?
if his goal is a raskin-like system then it would be better to simply begin with Linux (or any other OS for that matter) and build a raskin-like GUI on top of the frame buffer.
the object system described (objects within containers with attributes) sounds like a hierarchical file system (or an extension thereof) to me.
the question is not “what would we do if we could re-write linux“, but “what would we want to do that requires changes to linux“.
…will not involve so much software as it will hardware.
we’re running on hardware that’s designed to be backwards compatible with 20 year old technology. obviously, software follows suit…and you get crappy shit like Windows and Linux, with a keyboard, point-click interface, and almost no standards
the article is probably relevant for the next 3 or 5 years. okay, now designing and “publishing” new hardware requires more resources than writing new software. I’m writing a new OS myself, Cefarix, and taking into account all these points…but the single most frustrating thing is the hardware. only yesterday i got my radeon igp320m chip in my presario 2100 laptop partly working with 3d acceleration under X and Linux, by patching stuff in the server and the kernel. what about those people who dont know how to code? why cant we have any uniformity in the way hardware is accessed.
IBM, Intel, and Microsoft have really screwed us up IMHO. if they had followed set standards and principles, keeping in mind usability and productivity, we’d be lightyears ahead with computers today.
the problem isnt the software, nor the hardware, but bureaucracy and stuff. bloat comes from bureaucracy, to hardware, to software, and finally into people’s brains…which is why so many people are dumb nowadays. we need get rid of the backwards compatibility thing, we’re living in the past. we need to live in the future. as for the article, it’s just a repetition of the past, not even the present.
but the single most frustrating thing is the hardware. only yesterday i got my radeon igp320m chip in my presario 2100 laptop partly working with 3d acceleration under X and Linux, by patching stuff in the server and the kernel. what about those people who dont know how to code? why cant we have any uniformity in the way hardware is accessed.
Good point, I want my stuff to be simple to setup.