Slashdot Mirror
QNX: When an OS Really, Really Has to Work
An anonymous reader writes "Fortune has this article about how QNX's OS has found a niche and is doing well. Especially after 1996 when Microsoft executives said they would crush them in 2 years. When your software absolutely positively needs to work!"
You really are a fuckin' idiot aren't you?
Do you even know what an embedded system is, assgerbil?
Monolithic kernels, despite their theoretical inferiority, will be faster and more efficient because more developers will be working on them, and will be able to resolve inefficiencies faster.
But that's only initially, when they are small and simple. Once they have become big and complex, they get bogged down.
The Linux kernel is at that stage: kernel builds are a complete mess, as is adding and integrating new functionality. Every Linux machine I ever install the kernel on requires a recompilation to make all the hardware accessible, and then it takes hours of fiddling to find just the right combination of options to make it all work. And that's the experience of pretty much every other Linux user I know. Even then, things don't work properly and regress in serious ways (e.g., the 2.4.20 kernel corrupts file systems mounted via USB2, something that is working in 2.4.19). You can also see the problems with Linux kernel development in the mess around Bitkeeper--if the kernel were architected better, you wouldn't need a very high-tech source code management system to deal with it.
The Linux kernel will collapse under its own weight sooner or later: it simply isn't architected to be extensible.
So, the problems with monolithic archictectures are far from "theoretical", they are concrete and we are living with them every day. And they don't get fixed--there aren't enough programmers around to fix them. And, in fact, mature monolithic projects end up having a small fraction of the programmers they could have if the same projects were designed for extensibility and modularity.