[
https://issues.apache.org/jira/browse/HDFS-5650?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13844907#comment-13844907
]
Chris Nauroth commented on HDFS-5650:
-------------------------------------
bq. The v2 patch flatten the Acl class into the AclStatus class, which makes it
more amenable for serialization.
This means that we no longer have a class suitable for use in the NameNode
implementation, where we wanted to reuse {{Acl}} instances between multiple
files. ({{AclStatus}} includes the owner/group of one specific file, so it
isn't suitable.)
Are you suggesting that we make {{Acl}} internal to HDFS? We could do that.
The implementation would be somewhat duplicated (basically {{AclStatus}} minus
> Remove AclReadFlag and AclWriteFlag in FileSystem API
> -----------------------------------------------------
>
> Key: HDFS-5650
> URL: https://issues.apache.org/jira/browse/HDFS-5650
> Project: Hadoop HDFS
> Issue Type: Sub-task
> Components: hdfs-client, namenode, security
> Reporter: Haohui Mai
> Assignee: Haohui Mai
> Attachments: HDFS-5650.000.patch, HDFS-5650.001.patch,
> HDFS-5650.002.patch, HDFS-5650.003.patch
>
>
> AclReadFlag and AclWriteFlag intended to capture various options used in
> getfacl and setfacl. These options determine whether the tool should traverse
> the filesystem recursively, follow the symlink, etc., but they are not part
> of the core ACLs abstractions.
> The client program has more information and more flexibility to implement
> these options. This jira proposes to remove these flags to simplify the APIs.
--
This message was sent by Atlassian JIRA
(v6.1.4#6159)
