You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@thrift.apache.org by "David Mollitor (Jira)" <ji...@apache.org> on 2020/10/02 19:10:00 UTC

[jira] [Created] (THRIFT-5288) Better Support for ByteBuffer in Compact Protocol

David Mollitor created THRIFT-5288:
--------------------------------------

             Summary: Better Support for ByteBuffer in Compact Protocol
                 Key: THRIFT-5288
                 URL: https://issues.apache.org/jira/browse/THRIFT-5288
             Project: Thrift
          Issue Type: Improvement
            Reporter: David Mollitor
            Assignee: David Mollitor


{code:java|title=TCompactProtocol.java}
  /**
   * Write a byte array, using a varint for the size.
   */
  public void writeBinary(ByteBuffer bin) throws TException {
    int length = bin.limit() - bin.position();
    writeBinary(bin.array(), bin.position() + bin.arrayOffset(), length);
  }
{code}

I was working on something with Parquet and this code was causing some issues:

{code}
java.lang.Exception: java.nio.ReadOnlyBufferException
	at org.apache.hadoop.mapred.LocalJobRunner$Job.runTasks(LocalJobRunner.java:462)
	at org.apache.hadoop.mapred.LocalJobRunner$Job.run(LocalJobRunner.java:522)
Caused by: java.nio.ReadOnlyBufferException
	at java.nio.ByteBuffer.array(ByteBuffer.java:996)
	at shaded.parquet.org.apache.thrift.protocol.TCompactProtocol.writeBinary(TCompactProtocol.java:375)
	at org.apache.parquet.format.InterningProtocol.writeBinary(InterningProtocol.java:135)
	at org.apache.parquet.format.ColumnIndex$ColumnIndexStandardScheme.write(ColumnIndex.java:945)
	at org.apache.parquet.format.ColumnIndex$ColumnIndexStandardScheme.write(ColumnIndex.java:820)
	at org.apache.parquet.format.ColumnIndex.write(ColumnIndex.java:728)
	at org.apache.parquet.format.Util.write(Util.java:372)
	at org.apache.parquet.format.Util.writeColumnIndex(Util.java:69)
	at org.apache.parquet.hadoop.ParquetFileWriter.serializeColumnIndexes(ParquetFileWriter.java:1087)
	at org.apache.parquet.hadoop.ParquetFileWriter.end(ParquetFileWriter.java:1050)
{code}

This happens, because not all {{Buffer}}s allow for direct access to the backing Array,... for example a ByteBuffer tied to a file does not have an Array.  Read-only (immutable) {{ByteBuffer}s do not allow for this kind of access to the array since it could then be modified.

There are two approaches here:

# Assert and throw Exception if the backing array must be allowed for access
# Make "deal directly" with the ByteBuffer

I propose the latter.  However, the initial naive I approach I propose is to "deal directly" with the ByteBuffer by making a copy of the contents.



--
This message was sent by Atlassian Jira
(v8.3.4#803005)