[Cryptography] Do capabilities work? Do ACLs work?

[Bill Frantz]

On 2/10/15 at 9:54 AM, nico at cryptonector.com (Nico Williams) wrote:

>Capabilities in the academic (i.e., not Linux) sense are great, but we
>don't have enough real-world experience with them because, as you say,
>ACLs won out.

There is a bit of large scale experience with what I like to 
call web keys. These are unguessable URLs which provide access 
to a resource. As far as I can tell, DropBox uses them for 
files. When I put a bunch of photos into my DropBox and then 
send the URL to someone else, they can read the files with no 
user ID, just the URL.

The stated goal of Javascript/ECMAscript standardization is to 
move it to being a capability language. This direction isn't 
much of a stretch since memory-safe languages with some kind of 
object support already have a very strong capability base. You 
can't access an object without a reference, and the only way you 
can get a reference is to either create the object or to have 
the reference passed to you. Removing ways to get references 
which go around these two mechanisms is how you take an object 
language and make it a capability language.

Cheers - Bill

-----------------------------------------------------------------------
Bill Frantz        |The nice thing about standards| Periwinkle
(408)356-8506      |is there are so many to choose| 16345 
Englewood Ave
www.pwpconsult.com |from.   - Andrew Tanenbaum    | Los Gatos, 
CA 95032