[ 
https://issues.apache.org/jira/browse/SOLR-1549?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13445986#comment-13445986
 ] 

James Dyer commented on SOLR-1549:
----------------------------------

The patch looks simple and straightforward but I question the wisdom of 
allowing anything else but "abort" in the case of an SqlException.  You can't 
simply continue and read the next row of the query when this happens.  My 
thinking is "onAbort" is more for handling whole text documents that for some 
reason don't parse correctly.

I guess if it was a child entity doing n+1 selects it could just go do the next 
select with the following document.  Perhaps if the parent entity sometimes 
passes it a wrongly-typed join value this would happen and it would be ok to go 
on, assuming the next doc would have a correct join value?  But then again, you 
could just craft your sql to handle this contingency in these cases.

So if Sascha, Alexander or anyone can explain better why this is a good idea, 
and possibly contribute a unit test then perhaps we can commit this.  Otherwise 
I think its a "won't fix".
                
> SqlEntityProcessor does not recognize onError attribute
> -------------------------------------------------------
>
>                 Key: SOLR-1549
>                 URL: https://issues.apache.org/jira/browse/SOLR-1549
>             Project: Solr
>          Issue Type: Improvement
>          Components: contrib - DataImportHandler
>    Affects Versions: 1.4, 1.5
>            Reporter: Sascha Szott
>            Assignee: James Dyer
>            Priority: Minor
>         Attachments: SOLR-1549.patch
>
>
> Unfortunately, the SqlEntityProcessor does not recognize the value of an 
> entity's onError attribute in DIH's data config file. Therefore, in cases 
> where SQL exceptions are thrown somewhere inside the constructor of 
> ResultSetIterators (which is an inner class of JdbcDataSource), Solr's import 
> exits immediately, even though onError is set to continue or skip.
> In my opinion, use cases exist that will profit from database related 
> exception handling inside of Solr (e.g., in cases where the existence of 
> certain database tables or views is not predictable).

--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira

---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org
For additional commands, e-mail: dev-h...@lucene.apache.org

Reply via email to