Nick Allen created METRON-1062: ---------------------------------- Summary: SaltyRowKeyBuilder Handles Out of Order Timestamps Incorrectly Key: METRON-1062 URL: https://issues.apache.org/jira/browse/METRON-1062 Project: Metron Issue Type: Bug Reporter: Nick Allen Priority: Minor
In org/apache/metron/profiler/hbase/SaltyRowKeyBuilder... {code} * @return All of the row keys necessary to retrieve the profile measurements. */ @Override - public List<byte[]> rowKeys(String profile, String entity, List<Object> groups, long start, long end) { + public List<byte[]> encode(String profile, String entity, List<Object> groups, long start, long end) { // be forgiving of out-of-order start and end times; order is critical to this algorithm end = Math.max(start, end); start = Math.min(start, end); {code} Per [~mattf]: https://github.com/apache/metron/pull/622#discussion_r128650596 bq. Heh, this has been in the code for a long time, but isn't this a bug? If it starts out in the wrong order, say end is 1 and start is 5, won't this pair of statements result in both end and start being equal to the larger, ie 5 ? We need an intermediate variable for a binary swap! -- This message was sent by Atlassian JIRA (v6.4.14#64029)