[coyotos-dev] Hi + some stuff I found researching

[Haplo]

Because people are given deadlines so they can release their  
hardware, and would rather release their hardware sooner and make  
money and fix it later than they would to release their hardware with  
good, working drivers the first time =\. First they have to deal with  
people complaining about how long it takes, then they have to deal  
with people complaining that it sucks.

This apparently isn't the case ALL the time though. I've been running  
MacOSX for like 6 years now and never seen so much as a hint of a  
kernel panic. Not that I install 3rd party drivers all the time, but  
I can probably assert that the drivers included with the OS have  
stricter requirements than others. Maybe Apple has better policies  
for that, maybe they provide better support and seminars for doing it  
the right way, maybe more people who develop for mac go to WWDC than  
people who develop for windows go to.. whatever they might have, or  
maybe I'm just lucky, or don't install crappy drivers and don't buy  
crappy hardware. I don't really have enough of a test pool (woo, me  
and maybe 2 other people) to really say.

Point is, it really depends on company policy. Some applications will  
crash, others won't ever crash. Same with drivers. I'm starting to  
rethink my argument for a healing layer in any case, though. Seeing  
how I've never had a kernel panic it means that I've never installed  
a third party driver that can cause one. If only third party software  
is going to do that, in a microkernel system like coyotos it would  
only crash the related apps or disable the related hardware. This  
moves the blame from "YOUR OS HAS KERNEL PANICS AND SUCKS" to "wow  
this hardware company is garbage" and inflicts a sort of unspoken  
expectation on those companies to push them to change their policy on  
software development and testing.

Sad how much work it requires to switch kernels though. It makes it  
unlikely that any big commercial OS developer will move in that  
direction.
Oh, and I seriously need to question where they get their research  
data from to conclude that hybrid kernels were ever a good idea.  
Especially ones based on mach's horrible IPC.

On Jul 13, 2006, at 6:17 PM, Christopher Nelson wrote:

>> Yes, there's no substitute for doing it right, but in the
>> real world people really do have deadlines and so they don't
>> have time to write self-restarting drivers or even produce
>> drivers that aren't going to crash.
>
> How come there's never time to do it right, but there's always time to
> do it over?
>
> _______________________________________________
> coyotos-dev mailing list
> coyotos-dev at smtp.coyotos.org
> http://www.coyotos.org/mailman/listinfo/coyotos-dev