[
https://issues.apache.org/jira/browse/HIVE-5928?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13850574#comment-13850574
]
Brock Noland commented on HIVE-5928:
------------------------------------
bq. interface HiveBaseAuthorizationProvider
bq. There will be a subclass of HiveBaseAuthorizationProvider
Since it doesn't look like we have implemented here...may I interject some
thoughts? I think we should start moving hive development from inheritance to
composition where possible[1]. This looks like a great place to start.
[1] http://en.wikipedia.org/wiki/Composition_over_inheritance
> Add a hive authorization plugin api that does not assume privileges needed
> --------------------------------------------------------------------------
>
> Key: HIVE-5928
> URL: https://issues.apache.org/jira/browse/HIVE-5928
> Project: Hive
> Issue Type: Sub-task
> Components: Authorization
> Reporter: Thejas M Nair
> Original Estimate: 120h
> Remaining Estimate: 120h
>
> The existing HiveAuthorizationProvider interface implementations can be used
> to support custom authorization models.
> But this interface limits the customization for these reasons -
> 1. It has assumptions about the privileges required for an action.
> 2. It does have not functions that you can implement for having custom ways
> of doing the actions of access control statements.
> This jira proposes a new interface HiveBaseAuthorizationProvider that does
> not make assumptions of the privileges required for the actions. The
> authorize() functions will be equivalent of authorize(<hive object>,
> <action>). It will also have functions that will be called from the access
> control statements.
> The current HiveAuthorizationProvider will continue to be supported for
> backward compatibility. There will be a subclass of
> HiveBaseAuthorizationProvider that executes actions using this interface.
--
This message was sent by Atlassian JIRA
(v6.1.4#6159)
