OSDev.org

MessiahAndrw wrote:I don't think there will be an ideal OS, in the same way there isn't an ideal way of cooking (Japanese, American West, Italian)

The difference is in a combination of flavors. But every pizza includes the same flour. Your OS alway will have memory manager and other stuff. And if you do not like the smell of a particular memory manger then you can change it for another with a better smell.

bluemoon wrote:You neglect the fact that everyone may has different goal, some are mutual exclusive. The direct consequence is everyone want a different OS.

Beside of the smell combination there are such factors as a wish to taste your own pizza or to mix it's vegetables by your own hands. But when you have your pizza and have a lot of vegetable mixing experience then you see the clear lack of funding for a really great pizza.

The proposed solution is here to circumvent the funding problem. But all osdevers who still want to play with a vegetable mixture, of course, should be given a chance to take such experience in greater details and with different flavors.

bluemoon wrote:OS is so complex that you can't really think it consist of only ONE architecture.

I really think it consists of only one essential set of components. And all other flavors can be added on top of the base pizza.

The Linux has grown as the mentioned pizza. And even it's monopolistic kernel was not a problem for the system growth. If there will be a kernel with easily switchable modules then we will have even less problems. And nobody stops us from discussing the architecture. It is the open issue and will be such for a very long time.

The Linux has grown as the mentioned pizza. And even it's monopolistic kernel was not a problem for the system growth.

Unless I am missing the joke here, I would hazard to say you meant 'monolithic', not 'monopolistic'.

<gratuitous-bashing target="microsoft">
If any system has a monopolistic kernel, it would be Windows 8, running in less than 4GiB of RAM. You'd be lucky to get Notepad to run under those circumstances, because of the loading from just the OS itself. Of course, that's Microsoft's fault, not for poor design so much as for defining the alleged system requirements by means of wishful thinking...
</gratuitous-bashing>

Schol-R-LEA wrote:

The Linux has grown as the mentioned pizza. And even it's monopolistic kernel was not a problem for the system growth.

Unless I am missing the joke here, I would hazard to say you meant 'monolithic', not 'monopolistic'.

Linux's kernel is open (in contrast to Windows), but it still is monopolistic. If you want to switch to some new memory manager or scheduler you will face a great problem - you should rewrite a lot of software pieces distributed along the whole kernel. It prevents you from changing the kernel on a scale of components like memory manager or scheduler. The components in the Linux kernel are tightly coupled and it makes the kernel to be the only possible choice, unless you write absolutely new kernel from scratch. Isn't it a monopoly?

embryo wrote:

Schol-R-LEA wrote:

The Linux has grown as the mentioned pizza. And even it's monopolistic kernel was not a problem for the system growth.

Unless I am missing the joke here, I would hazard to say you meant 'monolithic', not 'monopolistic'.

Linux's kernel is open (in contrast to Windows), but it still is monopolistic. If you want to switch to some new memory manager or scheduler you will face a great problem - you should rewrite a lot of software pieces distributed along the whole kernel. It prevents you from changing the kernel on a scale of components like memory manager or scheduler. The components in the Linux kernel are tightly coupled and it makes the kernel to be the only possible choice, unless you write absolutely new kernel from scratch. Isn't it a monopoly?

You couldn't have picked worse examples. There are at least three each of different schedulers and memory managers inside the Linux kernel.

Owen wrote:There are at least three each of different schedulers and memory managers inside the Linux kernel.

Can you provide some links?

embryo wrote:

Owen wrote:There are at least three each of different schedulers and memory managers inside the Linux kernel.

Can you provide some links?

http://lwn.net/Articles/311502/

This article describes the addition of a fourth slab allocator for the Linux kernel. I believe the one described was never fully upstreamed, but there are still three that you can choose from when you configure a Linux kernel. This is because (despite your claims), Linux has a well defined interface that a slab allocator should implement.

dschatz wrote:http://lwn.net/Articles/311502/

This article describes the addition of a fourth slab allocator for the Linux kernel.

Yes, I should agree, there are some components and Linux supports component variation.