[e-lang] When will we see a solid capability OS?

[John Carlson]

>
>> Some think that we should ignore the OS and focus on the networking
>> aspects.
>
> Hm, or I guess on a programming language? It would presumably be
> easier to write a capability OS in a capability-oriented language?

I don't know how amenable programming languages are to capabilities.   
What I am thinking is that most bytecode these days is subject to  
weaving.  I'm sure that people are even thinking about weaving machine  
code.  Thus you must control the environment the code is running in.
(programs must be given the capabilities they have authority for from  
the user environment--or via the user's identity in Jed's approach).   
That said, you can add capabilities to an existing system through  
weaving, as well as snooping on capabilities.

Is anyone working on capability aspects?  Can you programmatically or  
declaratively remove unfriendly aspects?

If software is subject to change, then we have to rely on the  
hardware.  But with Albert and the COLA message passing languages, I'm  
starting to doubt our capability to protect the hardware.  We have to  
make an environment where the user is knowledgeable enough to make the  
right decisions, and can be an active participant in protecting the  
hardware.

See:
http://www.piumarta.com/papers/albert.pdf

John