Mutable ACLs are hard. Static ACLs are restrictive. Is there some middle ground?
One straightforward way to implement a static ACL is to use some part of the thread ID to encode the ACL's address, enabling all participants to independently find the AC rules.
A small improvement here: Instead of using a single ACL, use a chain of ACLs, where each one points to the previous. Some part of thread ID would then point to the (or one of the) ACL HEADs. This has a couple nice features:
Anyone can sort of "mutate" the ACL by forking the thread
I really really like this proposal a lot. It is simple, but 'fixes' a lot of complex issues. I think we should discuss this as a core idea in our call with @aschmahmann this week.
Mutable ACLs are hard. Static ACLs are restrictive. Is there some middle ground?
One straightforward way to implement a static ACL is to use some part of the thread
ID
to encode the ACL's address, enabling all participants to independently find the AC rules.A small improvement here: Instead of using a single ACL, use a chain of ACLs, where each one points to the previous. Some part of thread
ID
would then point to the (or one of the) ACLHEAD
s. This has a couple nice features: