[cap-talk] Wikipedia: Object-capability model - reference vs. capability?

[Norman Hardy]

On Jan 17, 2007, at 2:46 PM, Bill Frantz wrote:

> erights at gmail.com (Mark Miller) on Wednesday, January 17, 2007 wrote:
>
>> In KeyKOS, my understanding is that two independently created number
>> keys representing the number 7 cannot be distinguished. Number keys
>> are compared only on their contents, not on whether they were created
>> by the same act of creation. Is this correct?
>
> The manual is not clear on this issue, but the only handle the kernel
> has on "sameness" for data keys is the bits in the key.  If DISCRIM is
> to return true for some data key comparisons and false for others,  
> then
> it has no choice but to compare the bits of the keys.

My firm opinion is that the operation to produce the number key with  
7, always produces the same key, DK(7) in Keykos jargon.
It corresponds exactly to the common way of speaking where a floating  
point add instruction adding 3. and 4. always produces the same  
floating point value, namely 7. A key is a value, but is not data.

The operation on a page key to produce a RO page key, issued to the  
same page key, always produces the same RO key.
There is only one RO page key to that page. It existed before anyone  
put it in a slot, just like the floating point value  
3.288549890937749 exists whether or not any currently running  
computer is using that particular number just now.
When you ask if two floating numbers are the same, you don't ask when  
they were computed!
This corresponds to how the actual bits in the capability work, this  
persoective is not even smoke and mirrors.