The trick there is to use a solid code base that then encapsulates and
sand-boxes the other non-native OSs. This is particulary beneficial
for DOS and Win9x, as these need their view of the system speed and
capacities to be less than it really is - so the performance impact of
the emulation overhead is not a problem.
In 2006, you should really see DOS, Win9x, and even Win9x as
non-native with respect to today's hardware. This will become even
more of a factor as 64-bit, EFI, no-execute etc. take hold; it's
already demanded by USB, >137G, large RAM, fast CPU clock speeds,
altered relative timings for different CPU instructions, and less
attention paid to legacy BIOS standards.
Sorry, I think it's a terrible idea. Microsoft's security problems are in
part due to the time, trouble and money it costs them to support so many
different software variations. The customers are much better off if
Microsoft picks one code base and runs with it.
This is potentially true. Alas, real-world mileage has been poor
because MS doesn't grasp how different are the needs outside of NT's
non-traditional market, or they undervalue the importance of these.
The problems consumers have with XP are not because it's a pure Win432
code base that doesn't properly support Win9x, DOS and Win3.yuk apps.
Instead, it's because the XP use is unchanged from the design
requirements of professionally-administered network computing.
Windows 98 is only more secure if you focus on just one very narrow
definition of security... and a new release of Win98 with RPC/DCOM
and other things added, who knows how secure that might be.
I'm with Karl on this one. Win9x is safer only because there is less
of it - and especially because it doesn't open itself up to be used as
a network chew-toy, as NT is designed to do.
Once you port those mistakes into Win9x, you'd have all the un-safety
of the original XP plus all the insecurity of Win9x. What a mess!
Instead, how about rolling back NT to the bare-bones kernel, and then
applying the Win9x stand-alone design to delevoping it back up to a
full OS? IOW, none of that RPC, LSASS etc. and no facilities
whatsoever for any sort of remote admin. If you aren't physically at
the keyboard, you don't even have the right to speak to the OS unless
invited to do so by some outward-going traffic to your IP address, and
even thn, you don't have any admin access at all.
That gives you the safety of Win9x on the stability of NT, and uses a
common core code base for ease of support. The code base is better
not only because it's NT-based, but also because it's up to managing
modern hardware, in the same way that Win9x definitely is NOT.
I don't believe Win98 will make shared lab computers in public schools
more secure. A significant problem for such environments is insider
attacks and privilege escalation attacks. While XP is far from perfect
in this area, at least it tries; Win98 has zero defenses here.
Agreed. Kiosk PCs (i.e. those for use by a careless public) are hard
to manage, and while the simplicity of Win9x helps, it's not enough.
The main advantage of Win98 was the lower cost
And that evaporated when XP Home was released...
---------- ----- ---- --- -- - - - -
Proverbs Unscrolled #37
"Build it and they will come and break it"
