[Info-vax] FreeAXP loses network connectivity when laptop is woken up from "sleep"

[John Wallace]

On May 15, 1:55 pm, m.krae... at gsi.de (Michael Kraemer) wrote:
> In article <bed0d406-2f99-4d52-bfa1-26626f7dd... at z19g2000vbe.googlegroups.com>,
>
> John Wallace <johnwalla... at yahoo.co.uk> writes:
> > 2) The strange "file is there, file is not there" behaviour is a
> > feature introduced in Vista. It's related to apps that want to write
> > files to system-protected directories. When such a write happens in
> > Vista or later, it doesn't end up in the real system directory, there
> > is a shadow directory elsewhere that will get the file, that *some*
> > (but not all) APIs know about. It's a band-aid on an elastoplast to
> > work around a Defective by Design security model, but this is Windows.
>
> Wasn't Windows designed by Himself (aka D.Cutler)?

Not correct, even if you were referring specifically to the NT flavour
of Windows.

WNT's kernel stuff, process architecture, etc has some VAXELN
heritage. VAXELN was a not particularly well known (even inside DEC)
Cutler project for a distributable RT OS which would feel comfortable
for VMS programmers without being VMS, and allow distributed RT
applications to be developed without need to understand low level
hardware specifics and OS kernel interface details. VAXELN
incorporated early examples of a process model which also incorporated
threads, and a nice approach to interprocess data sharing (a
distributed naming service, transparent messaging between apps whether
on the same node or separate, etc). Marvellous stuff, some of which
duly made its way into NT, though many writers understandably missed
the VAXELN connection (it is briefly mentioned in Custer's book
"Inside Windows NT").

Unfortunately, Gates didn't like the fact that the robustness and
security made possible by things like the multiple protected address
spaces in VAXELN (and in any other worthwhile OS) meant that for some
benchmarks, WNT was slower than W98 on the same hardware (because of
things like the increased number of context switches, mode changes
etc). Productivity goes up as robustness goes up but isn't so easy to
demonstrate as frame rates on a game or other silly benchmark. So
Gates got the team to move quite a bit of code from user mode to
kernel mode, processes started using shared memory rather than
messaging, etc. Thereby increasing performance but inevitably also
decreasing robustness, as accidents were no longer confined to a
single process.

That's one aspect relevant here.

Another aspect of the defectiveness by design referred to earlier is
that Gates wanted the new OS to provide transparent compatibility with
existing applications but seemingly didn't understand the security
implications of that (why would he care anyway?). Consequently, random
apps were allowed to write pretty much where they wanted, because
that's what they'd always been able to do on Windows. Applications
couldn't do that on RSX or VMS or even VAXELN, OSes which all had some
concept of security.

Windows had previously had no concept of security, and therefore, for
compatibility with that legacy, neither did the Gatesified WNT.

WNT was not, is not, and never will be VMS++.