On Mon, 2007-06-04 at 13:55 +0530, Vihan Pandey wrote:
Really? Was it not you that said ``the difference between Andrew Tanenbaum and his "theories" and Linus who built a practical kernel - which works!". Try getting Operating System Design and Implementation(which comes with a Minix CD) then start hacking around with the exercises.
I'd rather learn with Linux than with Minix.
Besides, IIRC, microkernels
are plagued by the classic problem of "message passing" - too much message passing - which makes it, well, SLOW. Hello? indeed, and do you know the minuscule kernel zkernel achieved by it
the very idea is to have a minuscule kernel and rest of the crap in user space which is communicated with by message passing. The result is if any of the crap crashes, the kernel still survives.
*yawn* try telling me something new :)
We dont want to
follow the M$ ideology:
Hardware speed * Software bloat = constant
In fact the very idea of microkernels opposes all of M$'s design concepts(if they ever had any).
IIRC, M$'s own kernel ( Win2k or Win2k3 ) resembles a microkernel. yes, they too are apparent aware of the positive aspects of the microkernel.
Yes, I have studied quite a bit of Computer Science. And yes, I am graduating from college with flying colours. Thank you very much.
Good for you, now when you enter the real world you will see how much meaning the flying colours have against logic and reason.
I know. I've been in the "real world" for more than 4 years now :) I've seen both sides of the spectrum for quite some time..! ;)
A design that only solves short term problems eventually dies and takes all dependent on it to the grave.
Give me one microkernel replacement for the Linux kernel which you can substitute at this moment and build an entire "Free" operating system like GNU/Linux. Obviously, its performance should atleast match upto GNU/Linux.
Linux is a solution but by no means for the long term.
I never said Linux was a silver bullet. I never said it was a long term solution. You're putting words in my mouth.
The only trivialize that has been done here is by you of AST which IMHO was totally uncalled for. Moreover, dismissing things just because they don't work is a very nearsighted approach good only for companies with a short life span and not the community at large.
I've _not_ trivialized Andrew Tanenbaum at all. His work and his theories are respected around the world. Linus' creation has been around for some time now and is also respected in a similar way. The way people talk about microkernels being the salvation of the modern operating systems drives me up the wall.
Its a great concept but one that is plagued by implementation problems given todays technology.
IIRC, even RMS acknowledges that microkernels are tough to build and the GNU project's efforts in building one was taking waaay too long and around that time Linus came up with his kernel which the community accepted with open arms. ( Refer Revolution OS :P )