Karl Wettin (JIRA) wrote:
[ https://issues.apache.org/jira/browse/MAHOUT-25?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12590577#action_12590577 ]
Karl Wettin commented on MAHOUT-25:
-----------------------------------

+1, patch looks good.

Somewhat related, I wish AbstractMatrix.ROW and COL was an enumeration instead 
so it could be import it to other classes where you then only have to type ROW 
rather than AbstractMatrix.ROW.

But perhaps that would force me to type ROW.intValue() instead. Think I still 
prefere that.

Minor bugs/issues from code inspection
--------------------------------------

                Key: MAHOUT-25
                URL: https://issues.apache.org/jira/browse/MAHOUT-25
            Project: Mahout
         Issue Type: Bug
         Components: Matrix
   Affects Versions: 0.1
        Environment: All
           Reporter: Sean Owen
           Priority: Minor
            Fix For: 0.1

        Attachments: MAHOUT-25.patch


Hi all, just started checking out the code base to get familiar with it and 
turned loose IntelliJ on the code. It picked up a few possible issues I thought 
I'd highlight:
MatrixView:57
      for (int col = offset[COL]; col < offset[COL] + cardinality[COL]; row++)
Pretty sure that should be col++ at the end.
DenseMatrix:122
Instances are compared uisng == instead of equals(). Doesn't matter if this is 
exactly the semantics you want, but if DenseMatrix ever defined a notion of 
equals() this wouldn't use it and might be a bug. Same in many other classes.
Canopy:146, 168
      pointStronglyBound = pointStronglyBound | (dist < t2);
Should this really be a non-shortcircuit, versus
       pointStronglyBound = pointStronglyBound || (dist < t2);
or just
       if (!pointStronglyBound) {
              pointStronglyBound = dist < t2;
       }
CanopyDriver:52,53
String.valueOf(x) is a smidge faster than "" + x.
Actually I've got several more but they're less important, like redundant casts 
or utility classes without private constructors, etc. I can keep going here... 
want to help polish a few things but don't want to get overbearing.

+1 on this and the other changes too. I just committed a test for MatrixView and some MatrixView changes from MAHOUT-23 that may have corrected one problem you found, so update first as there may be one conflict.

Jeff

Reply via email to