[ 
https://issues.apache.org/jira/browse/ACCUMULO-1006?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13565777#comment-13565777
 ] 

Christopher Tubbs commented on ACCUMULO-1006:
---------------------------------------------

There's an implication here that I'm concerned about: we can catch and filter 
bad visibility labels during major compaction, if a file got constructed 
improperly in older AccumuloFileOutputFormat code. However, I'm not sure that's 
sure that's possible in the general case if the BadArgumentException does not 
get thrown and VisibilityParseException does. This is because the 
VisibilityEvaluator code short circuits the parse tree, depending on the 
authorizations passed to it. So, there may be cases where the visibility label 
is bad, passes the ColumnVisibility.validate(...), but fails to pass the 
VisibilityEvaluator, but we just don't see it because we didn't pass the right 
authorizations to trigger it. In such a case, we won't be able to filter those 
bad labels out. We need this test to see if this is a possible circumstance.
                
> Add test for VisibilityParseException
> -------------------------------------
>
>                 Key: ACCUMULO-1006
>                 URL: https://issues.apache.org/jira/browse/ACCUMULO-1006
>             Project: Accumulo
>          Issue Type: Bug
>          Components: tserver
>    Affects Versions: 1.5.0, 1.4.2
>            Reporter: Christopher Tubbs
>            Assignee: Keith Turner
>
> I'm not sure this exception can ever actually happen. I think 
> ColumnVisibility throws BadArgumentException in all the cases where the 
> VisibilityParseException would have been thrown. It's somewhat unfortunate 
> that the BadArgumentException is the RuntimeException, and the checked 
> exception can never happen.
> We need a test to see if we can make VisibilityParseException happen when 
> BadArgumentException doesn't, and possibly do something about 
> BadArgumentException, by either wrapping it with a checked exception, or 
> something else to reflect the fact that it's more likely to happen than 
> VisibilityParseException.

--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira

Reply via email to