[CLUE-Tech] (In)Stability of X?

[Michael J. Hammel]

Thus spoke Randy Arabie
> The nice thing about X is it can crash, but your box is still running. 
> IF Windows crashes, you've lost EVERYTHING, you gotta reboot.  I suppose 
> a Windows advocate would argue that the problem is not the OS, but the app. 
> Which may be true, you probably can run Windows with many days of uptime 
> if you don't run any apps.  I guess the protected memory improvements in 
> 2000 have made things better.  But not good enough (for the price), IMHO. 

This is the architectural difference between Windows and Unix.  X is not tied
into the kernel (unless you're using the newer framebuffer enabled servers).
So even if X dies, the OS doesn't.  Also, since applications are not tied
into X (they only communicate vie the X protocols), applications can die
without taking down the rest of your X environment.  (Framebuffer users are
on their own, however.)

On windows, because they've tied the display directly into the kernel and
thus tied applications there as well, a dying application can take down
everything.  The advantages for doing this are that things should be snappier
in response and apps have access to more low level functionality.  However,
I've not seen where this has made applications particularly more appealing
under Windows yet.  I get pretty snappy response under X and the architecture
of Unix/Linux makes access to devices from high level applications pretty
painless.

-- 
Michael J. Hammel           |
The Graphics Muse           |    I'm just working here till a good fast-food
mjhammel at graphics-muse.org  |    job opens up.
http://www.graphics-muse.com