It’s a bit of oversimplification, but a lot of people in the operating systems development community agree. Linux is a testament to manpower–not design genius.
You can’t just say linux design isn’t that great -and then just leave it at that.
Give me list of things that are fundamentally wrong or poorly designed about Linux.
-Linux was a hack but pretty much all the original code has been gutted and replaced by cleaner code over the years. Its system5 and posix compliant, its a good unix clone in my opinion and “hacked” parts of it get re-written -like the SMP scheduler. -The security system (root / user) is becoming more fine-grained. No doubt over time things will get stolen from plan9 and BSD -to become the ultimate bastard unix system.
The guys/girls from Mobius prob. got frustrated about
the fact that they thought they could do a better job than the so called Linux hackers but they found out the hard way, Linux is Damn good. Lets see where Mobius ends….
Its absence. All the design there is in Linux is the traditional Unix model – which is very powerful, but very nearly boils down to streams and pipes. That simplicity is an enormous asset as well as it is a limitation. In a deeper level, next to none of the novelties in kernel structure of the past 20 years have entered Linux. But hey, it’s based on a stout model, it grabs most of the best accessories that appear, it’s got tremendous support… it achieves real wonders given its design limitations.
LT would be the first to explain all that to you. In fact, I suspect he takes pride on having pushed Linux so far with so many limitations.
The implication is that Linux has no future. But neither do the others. It would be a sad future where we were bound to the currently available technology. It isn’t like it’s perfect.
You can’t just say linux design isn’t that great -and then just leave it at that.
Actually you can. It’s not like you have to write an essay on a subject everytime you refer to it. Were he to say ‘Linux’s design roolzz’, you woudn’t ask for clarifications.
The monolithic design has been replaced by a modular design which has evolved since its’ replacement, but a microkernel design cannot become true with Linux. You’ll have to start from scrath or use a microkernel with appropriate services then, like the GNU/Hurd. Before a war starts up, i’m not claiming either is a bad or good design. The developer ( Tim Robinson) chose a modular design as well, it is mentioned at the project goals.
Tim uses the name “DevFS”, while DevFS has a bad name and has been superseeded by Udev by now. I find this confusing.
Tim also state it is for IA32 only. Is this project’s code actually portable? This ain’t mentioned at the project goals.
while DevFS has a bad name and has been superseeded by Udev by now.
DevFS for Linux got a bad name because of politics. There was only one person working on it, and he disappeared. There were accusations that it had unsolvable problems, but only in relation to Linux inards. Other OSes have been using DevFS for a while now with success. I am personally disappointed in the drive to use udev, which is also implemented by a single developer and isn’t even complete yet. Udev also complicates the programmer’s task, because he may not know the device name to access without extra work. I consider this bad design.
There are more examples, especially surrounding power management. But more would be an article in itself. OS development involves tradeoffs. I like Linux for certain tasks, but not all.
What can’t be fixed about Linux?
Nothing, I suppose, given enough time and incompatibility. It all depends on whomever is driving the design.
This is an article about Moebius. Maybe we can move back to that now? I don’t know much about the OS itself, but Tim Robinson has written some good articles on OS development over at OSDever.net. They helped me write some specialized code for a recent project, and understand OSes in general better. So, hey, I’m pleased for the guy.
unix’s use the most successful design in the world that has been tested. people are starting to rival linux which shows that linux is a major player now
…and you have to enjoy their woefully mis-stated description of Linux.
“Linux has been going a long time, and has traditionally been a Unix clone.
As such it is a bit of a hack and is struggling to keep up with modern OS concepts.”
It’s a bit of oversimplification, but a lot of people in the operating systems development community agree. Linux is a testament to manpower–not design genius.
Whats wrong with Linux’s design?
You can’t just say linux design isn’t that great -and then just leave it at that.
Give me list of things that are fundamentally wrong or poorly designed about Linux.
-Linux was a hack but pretty much all the original code has been gutted and replaced by cleaner code over the years. Its system5 and posix compliant, its a good unix clone in my opinion and “hacked” parts of it get re-written -like the SMP scheduler. -The security system (root / user) is becoming more fine-grained. No doubt over time things will get stolen from plan9 and BSD -to become the ultimate bastard unix system.
What can’t be fixed about Linux?
The guys/girls from Mobius prob. got frustrated about
the fact that they thought they could do a better job than the so called Linux hackers but they found out the hard way, Linux is Damn good. Lets see where Mobius ends….
Whats wrong with Linux’s design?
Its absence. All the design there is in Linux is the traditional Unix model – which is very powerful, but very nearly boils down to streams and pipes. That simplicity is an enormous asset as well as it is a limitation. In a deeper level, next to none of the novelties in kernel structure of the past 20 years have entered Linux. But hey, it’s based on a stout model, it grabs most of the best accessories that appear, it’s got tremendous support… it achieves real wonders given its design limitations.
LT would be the first to explain all that to you. In fact, I suspect he takes pride on having pushed Linux so far with so many limitations.
The implication is that Linux has no future. But neither do the others. It would be a sad future where we were bound to the currently available technology. It isn’t like it’s perfect.
You can’t just say linux design isn’t that great -and then just leave it at that.
Actually you can. It’s not like you have to write an essay on a subject everytime you refer to it. Were he to say ‘Linux’s design roolzz’, you woudn’t ask for clarifications.
The monolithic design has been replaced by a modular design which has evolved since its’ replacement, but a microkernel design cannot become true with Linux. You’ll have to start from scrath or use a microkernel with appropriate services then, like the GNU/Hurd. Before a war starts up, i’m not claiming either is a bad or good design. The developer ( Tim Robinson) chose a modular design as well, it is mentioned at the project goals.
Tim uses the name “DevFS”, while DevFS has a bad name and has been superseeded by Udev by now. I find this confusing.
Tim also state it is for IA32 only. Is this project’s code actually portable? This ain’t mentioned at the project goals.
while DevFS has a bad name and has been superseeded by Udev by now.
DevFS for Linux got a bad name because of politics. There was only one person working on it, and he disappeared. There were accusations that it had unsolvable problems, but only in relation to Linux inards. Other OSes have been using DevFS for a while now with success. I am personally disappointed in the drive to use udev, which is also implemented by a single developer and isn’t even complete yet. Udev also complicates the programmer’s task, because he may not know the device name to access without extra work. I consider this bad design.
There are more examples, especially surrounding power management. But more would be an article in itself. OS development involves tradeoffs. I like Linux for certain tasks, but not all.
What can’t be fixed about Linux?
Nothing, I suppose, given enough time and incompatibility. It all depends on whomever is driving the design.
This is an article about Moebius. Maybe we can move back to that now? I don’t know much about the OS itself, but Tim Robinson has written some good articles on OS development over at OSDever.net. They helped me write some specialized code for a recent project, and understand OSes in general better. So, hey, I’m pleased for the guy.
on two systems I tried it on. I might file a bug report, but without VMWare it is a lot of bother to write down the dumps.
Ah well, such is the price of progress.
unix’s use the most successful design in the world that has been tested. people are starting to rival linux which shows that linux is a major player now