manifoldcf-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Karl Wright (JIRA)" <j...@apache.org>
Subject [jira] [Created] (CONNECTORS-993) Pipeline does not handle "no document" case properly
Date Wed, 16 Jul 2014 09:13:04 GMT
Karl Wright created CONNECTORS-993:
--------------------------------------

             Summary: Pipeline does not handle "no document" case properly
                 Key: CONNECTORS-993
                 URL: https://issues.apache.org/jira/browse/CONNECTORS-993
             Project: ManifoldCF
          Issue Type: Bug
          Components: Framework agents process
    Affects Versions: ManifoldCF 1.7
            Reporter: Karl Wright
            Assignee: Karl Wright
             Fix For: ManifoldCF 1.7


The incremental ingester and its pipeline code, and indeed transformation connectors themselves,
do not handle the "no document" case properly.

What should happen is that whenever a document is not retransmitted into a branch of the pipeline,
the document version should nevertheless be recorded in the ingeststatus table for the corresponding
output.  But transformation connectors can interfere with that transmission if they decide
simply to not index an incoming document.  They may also receive a null RepositoryDocument
into the transformation method, and will be unequipped to handle that.

Instead, I propose the following:
(1) An explicit IIncrementalIngester method needs to be invented for handling this case
(2) The activity class used for transformation methods should have a "noDocument()" method
(3) IncrementalIngester needs to propagate "noDocument" method calls down the pipeline without
calling transformations on them





--
This message was sent by Atlassian JIRA
(v6.2#6252)

Mime
View raw message