OS Comparison: QNX vs Linux

[Kevin Stallard]

Because I was wasting time, I found this site:

http://dataweek.co.za/news.asp?pklNewsI ... egoryID=37

There were a few things I added to my list of reasons why QNX is a better
choice than Linux in embedded systems.

Anyway...I posted this because I wanted to waste even more time

Cheerios,

Kevin

[Igor Kovalenko]

I would say most of that is very vague and some of it is plain wrong by
now. When you dig deep enough into the matter, there will be ultimately 2
things that are different:

1. QNX has much better latencies.
2. Linux has much better royalties.

The reality is that decisions are often made at the level where people
don't understand latencies talk, but they know royalties talk quite well.
And if the latencies are the problem? That's easy - throw RTAI into the
mix. Let them eggheads figure out how to make it all work. That's what we
pay them for, right?

-- igor

Kevin Stallard wrote:

Because I was wasting time, I found this site:

http://dataweek.co.za/news.asp?pklNewsI ... egoryID=37

There were a few things I added to my list of reasons why QNX is a better
choice than Linux in embedded systems.

Anyway...I posted this because I wanted to waste even more time

Cheerios,

Kevin

[Armin Steinhoff]

Kevin Stallard wrote:

Because I was wasting time, I found this site:

http://dataweek.co.za/news.asp?pklNewsI ... egoryID=37

There were a few things I added to my list of reasons why QNX is a better
choice than Linux in embedded systems.

Anyway...I posted this because I wanted to waste even more time

The statements about embedded Linux are simply un-qualified

For instance:

- recompilition of the Linux kernel isn't an issue, because you
have only to recompile the driver module for a modular kernel

- a driver module can be loaded and un-loaded at run-time

- QNX doesn't support more CPU families than Linux ... the
opposite is the case (e.g. Linux supports non-MMU CPUs)

- there are no GPL issues as long as you don't include GPL code
in your software

- a single programming failure in a kernel module doesn't crash
the whole system as long as the trap is correctly handled
(e.g. RTAI)

- a.s.o


True statements are e.g. :

- the Linux kernel is a fast moving target
(2.4xx -> 2.6.xx is completely different ...)

- the available docs, if there are any, are often outdated
(see above)

- the efforts to write kernel modules is therefore significant
higher compared with most commercial systems.

- that's also true for the real-time Linux'es


Regards

Armin Steinhoff

Cheerios,

Kevin