[ 
https://issues.apache.org/jira/browse/LUCENE-7189?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15231157#comment-15231157
 ] 

ASF subversion and git services commented on LUCENE-7189:
---------------------------------------------------------

Commit 07d99765f56265997aaa74d5e0eaa4401b667151 in lucene-solr's branch 
refs/heads/master from Mike McCandless
[ https://git-wip-us.apache.org/repos/asf?p=lucene-solr.git;h=07d9976 ]

LUCENE-7189: make it easier to write WebGL earth HTML for debugging geo failures


> Make BaseGeoPointTestCase more debuggable
> -----------------------------------------
>
>                 Key: LUCENE-7189
>                 URL: https://issues.apache.org/jira/browse/LUCENE-7189
>             Project: Lucene - Core
>          Issue Type: Bug
>            Reporter: Robert Muir
>
> This class tests queries (boxes, circles, polygons) against indexed documents 
> and fails if the query is wrong.
> Unfortunately it can be hard to debug (depending which test method failed, if 
> it was NIGHTLY, etc).
> It is making issues challenging/slower to debug for me on LUCENE-7185. In 
> general i have found debugging geo related issues is difficult, we should 
> invest more time to make this easier.
> I think we want to display:
> 1) original item (e.g. Rectangle, radius, polygon)
> 2) query.toString() <-- can be different in important ways!
> 3) how many total documents were wrong
> 4) info on up to N (say 5) wrong documents:
>   ID (not lucene docid) of doc.
>   Lat/Lon location(s) of doc
>   any relevant metric such as distance from origin
> 5) GeoTestUtil.toWebGLEarth output of html to visualize those N wrong 
> documents, the query itself, any additional stuff like bounding box that 
> might be useful plotted too.
> I think the wrong documents we should try for one-line (rather than many) and 
> we don't need to print stuff like deleted=false since these queries don't do 
> anything sneaky here around that.



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)

---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]

Reply via email to