Lex Spoon <lex at cc.gatech.edu> said:
>Unfortunately it's not that simple: you don't know when the individual
>parts of memory will get saved to disk, and so the on-disk version of
>the image will have different parts that are different ages.
>
It's not that simple to get it right and crash-proof, I agree. But I think
it'd be quite possible to get it proof-of-concept-proof, wouldn't it?

>Overall, EROS (and its predecossor KeyKOS) are to operating systems like
>Smalltalk is to languages: the core technology is decades old at this point,
>and yet they are still ahead of their time.
>
Well, I once read a paper talking about how new technologies in our business
don't get faster to market than in any other business (10-15 years), and it
seems to be correct...

>	http://www.cis.upenn.edu/~eros/
>
Thanks for the pointer. I looked at it some 1.5 years ago, when learning about
E - didn't do much with it back then, but at the moment with all the i386
virtual machines floating around, I might want to start experimenting with it.

-- 
Cees de Groot               http://www.cdegroot.com     <cg at cdegroot.com>
GnuPG 1024D/E0989E8B 0016 F679 F38D 5946 4ECD  1986 F303 937F E098 9E8B