Dylan Smith writes:

'Windows' is mainly a marketing name. There is a world of difference
between the Windows on GA GPS displays (it'll be the NT microkernel
running software designed by the display maker and will NOT contain the
Win32 subsystem nor the usual crufty device drivers) and Windows on a PC
(which does include the Win32 subsystem and a lot of other cruft).

Even the NT kernel has never been validated or certified for any
safety-of-life use that I'm aware of. There's just too much code.

It's the Win32 subsystem and various Windows drivers that people are
complaining about when they complain about the stability/quality of
Windows. Win32 itself is pretty vile - messy, looks like it was designed
by a dozen teams who never communicated with each other.

It received an infusion of the very poorly written code from Windows
95 when NT 4.0 was written, from what I recall. It made the OS more
friendly, but less stable and secure. But apparently Microsoft felt
that was what the market wanted, and they were probably right, at
least with respect to desktop machines.

Like car crashes, most Windows crashes are caused by bad drivers.

Virtually all, not just most.

A GA GPS display with the NT microkernel won't contain these buggy
drivers nor the hideous pile of cruft that is the Win32 subsystem. The
NT kernel itself is small and fairly elegant (even if the VMM, in my
opinion, leaves a lot to be desired). It's probably within the realm of
practicality to certify the NT kernel to a certain level of reliability.
(Whether it has been done or not I don't know).

Avionics software is sometimes verified instruction-by-instruction in
its final binary form. I have a hard time believing that anyone would
do that for NT, even for just the kernel. Even a Linux or UNIX kernel
would be hard to verify.

--
Transpose mxsmanic and gmail to reach me by e-mail.