Security models

Question

Security models

albe opened this issue 7 years ago · 2 comments

ACL based or claim based

https://goodenoughsoftware.net/2014/09/01/another-security-model/

Answer 1 · 2020-07-25T11:36:16.000Z

This should be added as a pluggable hook into the reading/writing process (preCommit, preFetch hooks), where the partition metadata is provided as argument. Userland code can then prevent the read/write from happening (or any other behaviour).

Answer 2 · 2020-07-25T11:46:26.000Z

Closing in favor of #126 for the reason mentioned above - I likely won't implement a full ACL system (would only be necessary for a cross-process API, e.g. http)