Re: Dedicated Z-wave sites? (driver comments)

["Robert Green" <ROBERT_GREEN1963@xxxxxxxxx>]

"Dennis Brothers" <brothers@xxxxxxxxxxxxxxxxxxxxxxx> wrote in message
news:4ueeepF17ekj2U1@xxxxxxxxxxxxxxxxxxxxx
> FYI, one of the advances touted for Windows Vista is a new driver model
> which pushes most of the driver code outside kernel space, making it more
> difficult for a badly-written driver to crash the OS.  The upside, of
> course, is that it should make crashes less likely.  The downside is that
> many drivers need to be rewritten, and some never will be - there's little
> incentive for manufacturers to support stuff that no longer contributes to
> their cashflow (the two-year-old Toshiba tablet PC I'm typing this on is a
> particularly depressing example :)

According to the NOOKs guy, that's a part of the solution - the part he
labels isolation.  I was surprised to learn how much kernel access drivers
have in both Windows and Linux.  He talked about drivers currently being
able to corrupt data structures in the kernel so severely there could be no
recovery.  Even with the monitoring program certain fatal errors were
unavoidable and usually indicated a very severe coding problem.

There were also some other pretty interesting modules, like recording input
in a spool file so that it can be "played back" if a driver has be
restarted.  A frank review of the typical driver creation process confirmed
what you've said as well.  There's little incentive to write these things in
the first place and there's almost none to support devices that are jnot
being sold anymore.  Ironically, this often leads to a lot of drivers being
written by earnest but inexperienced coders just to bring a device into
operation on a later OS than its "birth" OS.

--
Bobby G.