JoegenUSTC commented on issue #7132: URL: https://github.com/apache/gravitino/issues/7132#issuecomment-5276924983
Hi, I'm interested in the "column masking/pruning, column deletion" item listed under regulation supports in this EPIC. Before it lands, I'd like to understand the intended design direction: **1. Modeling: Policy vs access control?** Is column masking intended to be modeled as a **Policy** associated with a `COLUMN` metadata object, rather than being extended into access control? Today `Privilege.Condition` is strictly `ALLOW`/`DENY` (with DENY-wins), and `PolicyManager`'s supported object types don't yet include `COLUMN`. Confirming the intended home for masking rules would help. **2. Subject granularity — the key question.** Dynamic masking is usually **role/subject-granular** (e.g. Ranger, Snowflake, Databricks: the same column renders differently depending on the querying role). However, a Policy is associated with a *metadata object*, not with a subject (role/user). How is subject-granularity expected to be expressed? - Inside `PolicyContent` (e.g. a `role → maskType` mapping evaluated at query time), or - via a new policy-to-role/subject association mechanism? **3. Row filtering.** Same modeling question — is row-level filtering also planned to go through the policy system rather than access control? **4. Status & timeline.** Is there an existing sub-issue or design doc for masking/filtering we can follow, and any rough timeline? Thanks! -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected]
