sungwy commented on code in PR #13810:
URL: https://github.com/apache/iceberg/pull/13810#discussion_r2296566696


##########
open-api/rest-catalog-open-api.yaml:
##########
@@ -960,6 +960,15 @@ paths:
           schema:
             type: string
             enum: [ all, refs ]
+        - in: query
+          name: referenced-by
+          description:
+            The fully qualified name of the view (i.e namespace name and view 
name) that referenced the table.
+            
+            nested namespaces are separated by a unit separator (`0x1F`) byte.

Review Comment:
   Thanks for the thoughtful feedback, Prashant. I really appreciate you 
walking me through your reasoning :)
   
   You’ve raised some excellent points, and I agree with you on the following:
   - It’s not a great idea to introduce brand new concepts like “Trusted 
Engine” or “authorization” ad-hoc inside an OpenAPI parameter description.
   - The spec shouldn’t dictate how a catalog establishes trust, or which 
authorization mechanism is in use. That’s catalog-specific and rightly out of 
scope.
   
   That said, I do think there’s value in giving implementers clearer security 
guidance in the REST Catalog Spec to help them implement endpoints safely. Both 
your proposal here and the [FGAC 
RFC](https://docs.google.com/document/d/108Y0E8XsZi91x-UY0_aHLEbmXDNmxmS5BnDjunEKvTM/edit?tab=t.0#heading=h.87f6fmhmyr1n)
 lean heavily on the concept of trust between a client/engine and the catalog. 
For example, in the FGAC proposal, exposing row-filters or column masks 
securely depends on the assumption that the client is trusted to apply them 
faithfully and not leak security context. Without such a notion of trust, it 
becomes much harder to avoid threats like privilege escalation or 
confused-deputy problems.
   
   What are your thoughts on introducing the concept of “trust” in the [REST 
Catalog Spec documentation](https://iceberg.apache.org/rest-catalog-spec/)? The 
focus would be on defining the expected security traits of a trusted Iceberg 
client (e.g., that it will only use metadata and security context for their 
intended purpose, and not provide a general pathway for escalation), without 
prescribing how catalogs establish that trust. Once we have the concept agreed 
on by the community, perhaps we could reference it in the OPEN API yaml to 
provide a stronger security guideline.
   
   I'm curious to hear your thoughts on this suggestion. If it sounds 
reasonable, I'd be open to drafting a documentation PR, and starting a 
discussion on the mailing list to get feedback from the wider community.



-- 
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]


---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]

Reply via email to