[Bug 60005] debug versions of libraries

[Kuba Ober]

> A sane kernel would keep itself completely protected in its' own virtual
> machine you couldn't get at and would present you with a virtual machine
> which looked like you had complete freedom to do what you wanted.  All
> the real kernel would do is context switching. mapping of virtual hardware
> to real hardware, and provide some narrow form of message passing between
> virtual machines in order to allow bootstrapping and hardware allocation.
>  (Yes, I've used an OS which worked that way.  It was a great system for
> writing standalone code on.  You could write the code in a timeshared
> environment and actually have it work when you ported it to an empty
> machine.  Yes, you could bring up an instance of the OS in a virtual
> machine and it would think it had a real machine to run on -- if a bit
> slowly, and usually
> with fewer devices.)

I guess he'd be good to port VM/CMS over to i386, then. That would be oh sooo 
trivial. Hint: the CPU must allow recursive virtualization. i386 doesn't. 
That's why writing VMware is not a piece of cake. OTOH, on VM/CMS you get 
what VMware does for free, in hardware.

Seriously though, kdevelop-devel is no place to bicker about kernel and OS 
design :)

Cheers, Kuba Ober