accumulo-notifications mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "William Slacum (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (ACCUMULO-3718) not possible to create a Mutation object from scala w/o some extra helper code
Date Thu, 09 Apr 2015 04:58:13 GMT

    [ https://issues.apache.org/jira/browse/ACCUMULO-3718?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14486700#comment-14486700
] 

William Slacum commented on ACCUMULO-3718:
------------------------------------------

Yeah, I think we need a method like:

{code}
  private byte[] serializedSnapshot() {
    if (buffer != null) {
      return buffer.toArray();
    } else {
      return this.data;
    }
  }
{code}

and then we can use it in {{hashCode}} by modifying {{toThrift}}:

{code}
  public TMutation toThrift() {
    byte[] data = serializedSnapshot();
    return new TMutation(java.nio.ByteBuffer.wrap(row), java.nio.ByteBuffer.wrap(data), ByteBufferUtil.toByteBuffers(values),
entries);
  }
{code}

Note that {{equals}} also changes the state of the Mutation by calling {{Mutation#serialize}}.

> not possible to create a Mutation object from scala w/o some extra helper code
> ------------------------------------------------------------------------------
>
>                 Key: ACCUMULO-3718
>                 URL: https://issues.apache.org/jira/browse/ACCUMULO-3718
>             Project: Accumulo
>          Issue Type: Bug
>    Affects Versions: 1.5.2, 1.6.2
>            Reporter: David Holiday
>            Priority: Minor
>
> issue: 
> it's not possible to create a Mutation object from scala without employing a standalone
java jar wrapper. the preferred method for creating the object has you do it in two stages:
create with table row, then employ Mutation.put() to populate the object with the actual mutation
data. when you do this in scala, you get a
> java.lang.IllegalStateException: Can not add to mutation after serializing it at org.apache.accumulo.core.data.Mutation.put(Mutation.java:168)
at org.apache.accumulo.core.data.Mutation.put(Mutation.java:163) at org.apache.accumulo.core.data.Mutation.put(Mutation.java:211)
> error. I *think* this has something to do with the byte array going out of scope in Scala
but somehow not in Java. If you concat the operations (constuctor().put(data, data, ...) you
don't run into the error, but scala sees a Unit return type, so you can't actually add the
mutation to a BatchWriter. The only way I was able to get around this was to create a stand-alone
jar with a method that created then returned a populated mutation object. 
> I wasn't sure whether or not to call this a bug or an enhancement. given that you probably
want Accumulo to play nice with Scala I decided to call it a bug. 
> below is a link to the stack overflow thread I created whilst figuring all this out:

> http://stackoverflow.com/questions/29497547/odd-error-when-populating-accumulo-1-6-mutation-object-via-spark-notebook/29527189#29527189



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

Mime
View raw message