RFC for granting priviledges in snowflake

[ryanking]

I am trying to figure out the resource schemas for granting privileges in snowflake.

Below is a doc where I have some examples of how it could work. Please give feedback.

[codecov]

Codecov Report

Merging #13 into master will not change coverage.
The diff coverage is n/a.

@@           Coverage Diff           @@
##           master      #13   +/-   ##
=======================================
  Coverage   76.69%   76.69%           
=======================================
  Files           9        9           
  Lines         665      665           
=======================================
  Hits          510      510           
  Misses         93       93           
  Partials       62       62

[edulop91]

does this name do anything other than provide an ID to track tf state?

[ryanking]

Just state, because there is no natural ID otherwise.

[austinylin]

A few thoughts:

• Thanks for starting this!
• s/priviledges/privileges/g
• Just wanted to check my understanding, this is just how we assign privileges to roles? Users -> roles is another exercise?
• I think generally on the questions about how many permissions to enable to be nested in a single grant we should drive towards simplicity which to me would say let's make it easy to grant a ton of access in a single resource. I think that generally means we end up with one resource per role which is cognitively nice.

[ryanking]

• Thanks for starting this!
• s/priviledges/privileges/g

Yeah it seems i am unable to spell this work correctly.

• Just wanted to check my understanding, this is just how we assign privileges to roles? Users -> roles is another exercise?

Yes, there was already a PR and support for the later.

• I think generally on the questions about how many permissions to enable to be nested in a single grant we should drive towards simplicity which to me would say let's make it easy to grant a ton of access in a single resource. I think that generally means we end up with one resource per role which is cognitively nice.

ok