search

RubenVerborgh / solid-architecture

Drafts on the intended architecture of a concrete implementation of Solid
9 stars 3 forks source link

Change "authentication" to "credentials.Webid" #2

Closed TallTed closed 6 years ago

TallTed commented 6 years ago

"authentication" was being overused, to refer to both the process and a result of authentication. Now it only refers to the process, and "agentid" refers to the result.

RubenVerborgh commented 6 years ago

Agree. How about just webID though?

dmitrizagidulin commented 6 years ago

I'd propose we change it to credentials. WebID is not enough, the authentication result object also needs to contain various credentials & bearer tokens. (These are potentially used in authenticated fetches needed by some LDP operations (the fetching of group external acl files, and the fetching of the Source resource in a COPY operation).

TallTed commented 6 years ago

I think WebID (or webID) would suggest inappropriate limitation, and be prone to similar misunderstanding as authentication. credentials is not unreasonable, but discussion about what more might be here probably needs to go into an actual issue -- as it would also seem to change the value structure from a (simple) string to something more complex, and go beyond a simple editorial clarification of wording.

RubenVerborgh commented 6 years ago

A credentials object with a webID property?

dmitrizagidulin commented 6 years ago

A credentials object with a webID property?

Yeah!

TallTed commented 6 years ago

credentials.webID, credentials.mechanism, credentials.timestamp, credentials.expires, etc., may be going in the right direction.

authentication.webID, authentication.mechanism, authentication.timestamp, authentication.expires, etc., might work as well.

That all said, I repeat that the authentication process should not be referred to with the same (simple) lexical label as is used to refer to the output of that process (whether that becomes AgentID as I put in the PR, or credentials.webID, authentication.webID, or something else), and that's what this PR was and remains about. The changes in this PR would leave the doc in a better state for further revision based on the discussions here and on Gitter.

RubenVerborgh commented 6 years ago

That all said, I repeat that the authentication process should not be referred to with the same (simple) lexical label as is used to refer to the output of that proc

Indeed, I think we all agree on that.

The changes in this PR would leave the doc in a better state for further revision

Can you update to credentials.webID, or make your branch writable so I can?

melvincarvalho commented 6 years ago

Just a very minor note, can we be a bit careful on the capitalization of ID.

I think @dmitrizagidulin prefers WebId in order to be consistent with the code base. Just is a possible source of bugs that could take time to track down.

TallTed commented 6 years ago

@melvincarvalho, @RubenVerborgh - Capitalization concern noted; updated to use credentials.Webid

RubenVerborgh commented 6 years ago

I've taken this up in 81c92e66f9f7ed8ca78fc9fd3d6effde7598a9cb; I think the commits here were a bit too detailed. (No need to specify the exact name of the WebID field, this is just on a high level what bits of information need to be known by which components.)