[ 
https://issues.apache.org/jira/browse/HIVE-1529?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12898003#action_12898003
 ] 

Mayank Lahiri commented on HIVE-1529:
-------------------------------------

Hi Pierre,

The numerical results appear to be accurate. A couple of comments about the 
code:

(1) Run "ant checkstyle" and looks at the formatting errors for your file in 
the build/checkstyle/checkstyle-errors.html file. In particular, remove 
commented lines like #160 of GenericUDAFCovariance.java, and newline-elses like 
line #214, unnecessary wraps #210-211

(2) Is there any reason for accepting string arguments in the Resolver class? 
If the user has a numeric value as a string, they can simply (CAST val AS 
double) in the query. As it stands right now, passing junk strings as one of 
the input expressions causes a return value of NULL and a silent exception that 
is only visible in the log file. It might be better to simply not accept STRING 
types in the resolver, as in GenericUDAFHistogramNumeric.java. This would also 
mean that you don't have to test for a NumberFormatException in the iterate() 
method -- line #263 of GenericUDAFCovariance.java.

(3) Please add at least a little extended function info, line #59, see 
GenericUDAFHistogramNumeric.java or GenericUDAFnGrams.java for an example.

> Add ANSI SQL covariance aggregate functions: covar_pop and covar_samp.
> ----------------------------------------------------------------------
>
>                 Key: HIVE-1529
>                 URL: https://issues.apache.org/jira/browse/HIVE-1529
>             Project: Hadoop Hive
>          Issue Type: New Feature
>          Components: Query Processor
>    Affects Versions: 0.7.0
>            Reporter: Pierre Huyn
>            Assignee: Pierre Huyn
>             Fix For: 0.7.0
>
>         Attachments: HIVE-1529.1.patch
>
>   Original Estimate: 96h
>  Remaining Estimate: 96h
>
> Create new built-in aggregate functions covar_pop and covar_samp, functions 
> commonly used in statistical data analyses.

-- 
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.

Reply via email to