Permissions

Who may read and write, and why a suspended editor may not.

permissions.js · output · proof · check · try it in the playground

The question

Many systems give people roles: editors may read and write, viewers may only read. On top of that, someone can be suspended, and then they may do nothing.

Alice and Carol are editors, Bob is a viewer, and Carol is suspended. Who is allowed to do what?

What we tell eyemind

fact(role('alice', 'editor'))
fact(role('bob', 'viewer'))
fact(role('carol', 'editor'))
fact(permits('editor', 'read'))
fact(permits('editor', 'write'))
fact(permits('viewer', 'read'))
fact(suspended('carol'))
implies([role(User, Role), permits(Role, Action)], candidate(User, Action))
implies([candidate(User, Action), not(suspended(User))], allowed(User, Action))
query(allowed(User, Action))

First work out what each role would allow, then remove anyone suspended. not(...) means “nothing in the program says so”: a user counts as not suspended when there is no suspended fact for them.

What eyemind concludes

allowed('alice', 'read')
allowed('alice', 'write')
allowed('bob', 'read')

Carol is a candidate for reading and writing, but she is suspended, so she gets nothing.

Why: the proof in plain words

For Alice writing:

1. Alice is an editor — fact 1.
2. Editors may write — fact 5.
3. So Alice is a candidate to write — rule 8.
4. Alice is not suspended — searched, nothing found.
5. So Alice is allowed to write — rule 9.

Step 4 is different in kind: it is not a fact we gave, but the absence of one.

“Not suspended”: a claim about absence

eyemind works under a closed world: if the program does not say Alice is suspended, it takes her to be not suspended. That is what we want here — but it is a claim about what is missing from the data, and a proof cannot point at something missing.

So the checker does not pretend to prove it. It records it as an obligation.

Checked, not just claimed

• 11 steps are confirmed as instances of the program lines they cite.
• 2 steps are taken on trust as absent obligations: “Alice is not suspended” and “Bob is not suspended”. eyemind searched everything it knows and found nothing.
• The checker looked for evidence against both (a suspended('alice') fact, say) and found none.

Verdict: checked_with_obligations. 13 steps, 2 on trust.

Try it

eyemind examples/permissions.js
eyemind --check-proof examples/proof/permissions.js examples/permissions.js

Delete the line fact(suspended('carol')) and run again: Carol is now allowed to read and write, like Alice.

Takeaway

“Allowed unless excluded” decisions rest on what is not in the data. eyemind makes those absences visible, so you know exactly what you are trusting when you grant access.