[
https://issues.apache.org/jira/browse/ATLAS-5148?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=18066686#comment-18066686
]
Radhika Kundam commented on ATLAS-5148:
---------------------------------------
[~adityagupta36] ,
The 2.5.0 release is already in progress, and the branch has been closed for
any new commits. Since this Jira is not yet merged, we are removing 2.5.0 from
the Fix Version.
> Atlas DSL GROUP BY fails for non-String attributes (DATE/Long)
> --------------------------------------------------------------
>
> Key: ATLAS-5148
> URL: https://issues.apache.org/jira/browse/ATLAS-5148
> Project: Atlas
> Issue Type: Bug
> Reporter: Rahul Kurup
> Assignee: Aditya Gupta
> Priority: Major
> Fix For: 3.0.0, 2.5.0
>
> Time Spent: 20m
> Remaining Estimate: 0h
>
>
> {{Table groupby(createTime) select owner, name, max(createTime)}}
> I tried running this command in DSL search but it would return 0 results
> despite having data from quick_start.py
> However these subset queries worked
> {{Table groupby(createTime)}}
> {{Table select owner, name, max(createTime)}}
>
>
> *Problem Description (Before Fix)*
> +1. GROUP BY works only for String attributes+
> Query:
> {color:#ff0000}Table groupby(owner){color}
> Works
> Query:
> {color:#ff0000}Table groupby(createTime){color}
> Returns no records
> Both owner (String) and createTime (Date/Long) attributes exist on the same
> hive_table entities.
> +2. GROUP BY with SELECT fails for DATE attributes+
> Query:
> {color:#ff0000}Table groupby(createTime) select owner, name,
> max(createTime){color}
> Returns no records
> Similar query using a String attribute works as expected:
> {color:#ff0000}Table groupby(owner) select owner, count(){color}
> Works
> +3. Root Cause+
> GROUP BY keys were handled only for String data types
> When grouping by non-String attributes (Date/Long), valid group keys were
> dropped during result processing
> This resulted in empty results even though grouping succeeded internally
>
> *Fix Description (After Fix)*
> +1. Handle GROUP BY keys of any data type+
> Updated method:
> {color:#ff0000}AtlasJanusGraphTraversal.getAtlasVertexMap(){color}
> Logic change:
> {color:#ff0000}String keyStr = (key instanceof String) ? (String) key :
> String.valueOf(key);{color}
> This converts GROUP BY keys of type Long, Date, Integer, etc. into String
> Prevents valid GROUP BY results from being dropped
> +2. GROUP BY on DATE attributes now works correctly+
> Query:
> {color:#ff0000}Table groupby(createTime){color}
> Returns grouped results
> +3. DSL queries now behave as documented+
> The following queries now return valid results:
> {color:#ff0000}Table groupby(createTime) select owner, name,
> min(createTime){color}
> {color:#ff0000}Table groupby(createTime) select owner, name, max(createTime)
> {color}
> {color:#ff0000}
> {color:#172b4d}Doc Attached:{color}{color}
> {color:#ff0000}
> {color:#172b4d}https://docs.google.com/document/d/1fgbj38CQ0qQB3dc_Y5t8HE-tjaQMWo7O3XB0memhwaU/edit?usp=sharing{color}
> {color}
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
