[
https://issues.apache.org/jira/browse/HDFS-6375?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14011407#comment-14011407
]
Andrew Wang commented on HDFS-6375:
-----------------------------------
Upon looking closer, you're totally right. I think re-using the existing XAttr
serialization isn't going to work so well for WebHDFS also because of the
direct REST API; I was okay with the funny handling where we ignore the values
for the Java clients, since that's an internal implementation detail, but here,
the on-the-wire format is public. My bad.
Based on this, I think we should serialize a JSON array of just the string
names on the NN side, and then pull it out and return a plain List in
WebHDFSFileSystem. This way, REST clients get a nice format (no need to parse
the array of maps).
Does this sound reasonable to you?
> Listing extended attributes with the search permission
> ------------------------------------------------------
>
> Key: HDFS-6375
> URL: https://issues.apache.org/jira/browse/HDFS-6375
> Project: Hadoop HDFS
> Issue Type: Improvement
> Components: namenode
> Affects Versions: 3.0.0
> Reporter: Andrew Wang
> Assignee: Charles Lamb
> Attachments: HDFS-6375.1.patch, HDFS-6375.2.patch, HDFS-6375.3.patch,
> HDFS-6375.4.patch, HDFS-6375.5.patch, HDFS-6375.6.patch, HDFS-6375.7.patch,
> HDFS-6375.8.patch, HDFS-6375.9.patch
>
>
> From the attr(5) manpage:
> {noformat}
> Users with search access to a file or directory may retrieve a list of
> attribute names defined for that file or directory.
> {noformat}
> This is like doing {{getfattr}} without the {{-d}} flag, which we currently
> don't support.
--
This message was sent by Atlassian JIRA
(v6.2#6252)
