How would you add capabilities to Erlang without making a new language?

Question

I was wondering how someone would go about adding Capability-based security to Erlang without making it into a new language and preserving maximum backwards compatibility on the application side?

I was thinking parse transforms could be used to remove access to things like `erlang:processes()` as to not compromise the system without breaking system libraries. I was also thinking to make Pids unguessable for example something like `<0.48.202270464972395333011708598926446773344>` instead of `<0.48.0>` where `202270464972395333011708598926446773344` is a large random number. I have no idea if this makes sense to or not from a beam vm internals point of view. — Eric des Courtis, May 29 '13 at 13:08

score 6 · Accepted Answer · answered May 30 '13 at 07:24

6

If you really are interested in this you should read the Master's thesis by Gustaf Naeser on Safe Erlang

answered May 30 '13 at 07:24

Jan Henry Nystrom

1,045
5
8

Interesting +1 for that! – Eric des Courtis May 30 '13 at 12:55
Okay I guess this really answers my question. – Eric des Courtis May 30 '13 at 13:03

How would you add capabilities to Erlang without making a new language?

1 Answers1