-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/18917/#review36575
-----------------------------------------------------------


The patch should make this feature optional. Preferably, make the visibility 
evaluation pluggable. At the very least, there's good reasons to disallow NOT 
terms, so this shouldn't be the default.

I'd also like to see some additional tests, that read and write data to verify 
the feature end-to-end. This is changing some very important code that's been 
stable for some time, so more tests are always better.

- Christopher Tubbs


On March 7, 2014, 2:41 p.m., Joe Ferner wrote:
> 
> -----------------------------------------------------------
> This is an automatically generated e-mail. To reply, visit:
> https://reviews.apache.org/r/18917/
> -----------------------------------------------------------
> 
> (Updated March 7, 2014, 2:41 p.m.)
> 
> 
> Review request for accumulo.
> 
> 
> Repository: accumulo
> 
> 
> Description
> -------
> 
> This patch adds a NOT "!" operator to ColumnVisibility.
> 
> The syntax is as follows:
> !a
> (!a)&(!b)
> a&(!b)
> a&(!(b|c))
> 
> Because of the nature of the current visibility parsing algorithm the 
> additional parentheses are required.
> In the shell, the "Unable to render embedded object: File (" requires 
> escaping. This is due to how JLine parses the command and attempts to 
> substitute ") not found." with history.
> 
> 
> Diffs
> -----
> 
>   core/src/main/java/org/apache/accumulo/core/security/ColumnVisibility.java 
> 75091d2 
>   
> core/src/main/java/org/apache/accumulo/core/security/VisibilityEvaluator.java 
> 725b2c7 
>   
> core/src/test/java/org/apache/accumulo/core/security/ColumnVisibilityTest.java
>  7a6a80d 
>   
> core/src/test/java/org/apache/accumulo/core/security/VisibilityEvaluatorTest.java
>  ee4d2ee 
> 
> Diff: https://reviews.apache.org/r/18917/diff/
> 
> 
> Testing
> -------
> 
> This patch includes unit tests for parsing, flattening, and evaluating the 
> not operator.
> 
> 
> Thanks,
> 
> Joe Ferner
> 
>

Reply via email to