[
https://issues.apache.org/jira/browse/CONNECTORS-969?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14148502#comment-14148502
]
Karl Wright commented on CONNECTORS-969:
----------------------------------------
The two output connectors that still do it this way are HDFS and the file
output connector. In both cases, the root file path needs to be extracted from
the version string. But that is all that is needed.
> Output connectors that use a JSON description string likely to fail
> -------------------------------------------------------------------
>
> Key: CONNECTORS-969
> URL: https://issues.apache.org/jira/browse/CONNECTORS-969
> Project: ManifoldCF
> Issue Type: Bug
> Components: File system connector
> Affects Versions: ManifoldCF 1.6.1
> Reporter: Karl Wright
> Assignee: Karl Wright
> Fix For: ManifoldCF 2.0
>
>
> When JSON is used as a description string, the order of terms in JSON is not
> determined - it is hashmap dependent. This means that the strings cannot be
> compared reliably between different versions of Java. Since these strings
> are stored in the database, upgrading java versions for these connectors
> might well cause all records going to them to be reindexed.
> JSON is also quite wordy (although better than XML) and is not a great fit
> for database storage for this reason.
> The alternative is to use the pack primitives that are available from the
> base connector class.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
