You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@impala.apache.org by "Thomas Tauber-Marshall (JIRA)" <ji...@apache.org> on 2017/05/05 20:38:04 UTC

[jira] [Created] (IMPALA-5284) Eliminate our dependence of the Kudu Java client

Thomas Tauber-Marshall created IMPALA-5284:
----------------------------------------------

             Summary: Eliminate our dependence of the Kudu Java client
                 Key: IMPALA-5284
                 URL: https://issues.apache.org/jira/browse/IMPALA-5284
             Project: IMPALA
          Issue Type: Improvement
    Affects Versions: Impala 2.9.0
            Reporter: Thomas Tauber-Marshall
            Priority: Critical


Its difficult to ensure that we're always using the same versions of the Java and C++ Kudu clients, potentially leading to conflicts. It would be nice if we could eliminate our dependence on the Java client.

One way to do this would be to add JNI calls in the FE/catalog to call into the Kudu C++ client in the BE. Unfortunately, this could result in a lot of boiler plate code, as there are a lot of Kudu client functions that are called from the FE/catalog (e.g. tableExists, openTable, createTable, deleteTable, getSchema, getPartitionSchema, getColumns) that would need to be routed through the JNI.



--
This message was sent by Atlassian JIRA
(v6.3.15#6346)