You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@drill.apache.org by chunhui-shi <gi...@git.apache.org> on 2016/10/31 06:03:12 UTC

[GitHub] drill pull request #638: DRILL-4982: Separate Hive reader classes for differ...

GitHub user chunhui-shi opened a pull request:

    https://github.com/apache/drill/pull/638

    DRILL-4982: Separate Hive reader classes for different data formats t\u2026

    \u2026o improve performance.
    
    1, Separating Hive reader classes allows optimization to apply on different classes in optimized ways. This  separation effectively avoid the performance degradation of scan.
    
    2, Do not apply Skip footer/header mechanism on most Hive formats. This skip mechanism introduces extra checks on each incoming records.

You can merge this pull request into a Git repository by running:

    $ git pull https://github.com/chunhui-shi/drill DRILL-4982

Alternatively you can review and apply these changes as the patch at:

    https://github.com/apache/drill/pull/638.patch

To close this pull request, make a commit to your master/trunk branch
with (at least) the following in the commit message:

    This closes #638
    
----
commit 1e240a0dfea7970798bd550204388fc0e9bc3d42
Author: chunhui-shi <cs...@maprtech.com>
Date:   2016-10-30T08:29:06Z

    DRILL-4982: Separate Hive reader classes for different data formats to improve performance.
    
    1, Separating Hive reader classes allows optimization to apply on different classes in optimized ways. This  separation effectively avoid the performance degradation of scan.
    
    2, Do not apply Skip footer/header mechanism on most Hive formats. This skip mechanism introduces extra checks on each incoming records.

----


---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at infrastructure@apache.org or file a JIRA ticket
with INFRA.
---

[GitHub] drill pull request #638: DRILL-4982: Separate Hive reader classes for differ...

Posted by bitblender <gi...@git.apache.org>.
Github user bitblender commented on a diff in the pull request:

    https://github.com/apache/drill/pull/638#discussion_r90574793
  
    --- Diff: contrib/storage-hive/core/src/main/java/org/apache/drill/exec/store/hive/HiveAbstractReader.java ---
    @@ -218,17 +229,18 @@ private void init() throws ExecutionSetupException {
             }
           }
         } catch (Exception e) {
    -      throw new ExecutionSetupException("Failure while initializing HiveRecordReader: " + e.getMessage(), e);
    +      throw new ExecutionSetupException("Failure while initializing HiveOrcReader: " + e.getMessage(), e);
    --- End diff --
    
    "HiveOrcReader:" in the exception-string should be changed so that it prints the actual reader type. 


---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at infrastructure@apache.org or file a JIRA ticket
with INFRA.
---

[GitHub] drill pull request #638: DRILL-4982: Separate Hive reader classes for differ...

Posted by asfgit <gi...@git.apache.org>.
Github user asfgit closed the pull request at:

    https://github.com/apache/drill/pull/638


---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at infrastructure@apache.org or file a JIRA ticket
with INFRA.
---

[GitHub] drill issue #638: DRILL-4982: Separate Hive reader classes for different dat...

Posted by chunhui-shi <gi...@git.apache.org>.
Github user chunhui-shi commented on the issue:

    https://github.com/apache/drill/pull/638
  
    Addressed your comments.


---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at infrastructure@apache.org or file a JIRA ticket
with INFRA.
---

[GitHub] drill issue #638: DRILL-4982: Separate Hive reader classes for different dat...

Posted by chunhui-shi <gi...@git.apache.org>.
Github user chunhui-shi commented on the issue:

    https://github.com/apache/drill/pull/638
  
    The performance degradation is worse with larger tables. The degradation is >300% on a 15GB table. After this fix, this degradation disappears.
    
    There are other visible performance gains of this fix even when there is no degradation:
    
    For a simple query on ORC/Parquet table through HiveReaders, I observed the improvement is about 10%-25% and the average improved percentage is 17.6%.
    
    For TPCH tests on a 10 node cluster with Parquet tables + HiveReaders, the average improvement (of two runs) is 5.1%
    



---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at infrastructure@apache.org or file a JIRA ticket
with INFRA.
---

[GitHub] drill issue #638: DRILL-4982: Separate Hive reader classes for different dat...

Posted by parthchandra <gi...@git.apache.org>.
Github user parthchandra commented on the issue:

    https://github.com/apache/drill/pull/638
  
    LGTM too. +1


---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at infrastructure@apache.org or file a JIRA ticket
with INFRA.
---

[GitHub] drill pull request #638: DRILL-4982: Separate Hive reader classes for differ...

Posted by paul-rogers <gi...@git.apache.org>.
Github user paul-rogers commented on a diff in the pull request:

    https://github.com/apache/drill/pull/638#discussion_r91446474
  
    --- Diff: contrib/storage-hive/core/src/main/codegen/templates/HiveRecordReaders.java ---
    @@ -0,0 +1,300 @@
    +/**
    + * Licensed to the Apache Software Foundation (ASF) under one
    + * or more contributor license agreements.  See the NOTICE file
    + * distributed with this work for additional information
    + * regarding copyright ownership.  The ASF licenses this file
    + * to you under the Apache License, Version 2.0 (the
    + * "License"); you may not use this file except in compliance
    + * with the License.  You may obtain a copy of the License at
    + *
    + * http://www.apache.org/licenses/LICENSE-2.0
    + *
    + * Unless required by applicable law or agreed to in writing, software
    + * distributed under the License is distributed on an "AS IS" BASIS,
    + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
    + * See the License for the specific language governing permissions and
    + * limitations under the License.
    + */
    +
    +/**
    + * This template is used to generate different Hive record reader classes for different data formats
    + * to avoid JIT profile pullusion. These readers are derived from HiveAbstractReader which implements
    + * codes for init and setup stage, but the repeated - and performance critical part - next() method is
    + * separately implemented in the classes generated from this template. The internal SkipRecordReeader
    + * class is also separated as well due to the same reason.
    + *
    + * As to the performance gain with this change, please refer to:
    + * https://issues.apache.org/jira/browse/DRILL-4982
    + *
    + */
    +<@pp.dropOutputFile />
    +<#list hiveFormat.map as entry>
    +<@pp.changeOutputFile name="/org/apache/drill/exec/store/hive/Hive${entry.hiveReader}Reader.java" />
    +<#include "/@includes/license.ftl" />
    +
    +package org.apache.drill.exec.store.hive;
    +
    +import java.io.IOException;
    +import java.util.List;
    +import java.util.Properties;
    +import org.apache.drill.common.exceptions.DrillRuntimeException;
    +import org.apache.drill.common.exceptions.ExecutionSetupException;
    +import org.apache.drill.common.expression.SchemaPath;
    +import org.apache.drill.exec.ops.FragmentContext;
    +import org.apache.drill.exec.vector.AllocationHelper;
    +import org.apache.drill.exec.vector.ValueVector;
    +import org.apache.hadoop.hive.metastore.api.Partition;
    +import org.apache.hadoop.hive.metastore.api.Table;
    +import org.apache.hadoop.io.Writable;
    +import org.apache.hadoop.mapred.InputSplit;
    +import org.apache.hadoop.security.UserGroupInformation;
    +import org.apache.hadoop.hive.conf.HiveConf;
    +
    +import org.apache.hadoop.hive.serde2.SerDeException;
    +
    +import org.apache.hadoop.mapred.RecordReader;
    +<#if entry.hasHeaderFooter == true>
    +import org.apache.hadoop.hive.serde2.SerDe;
    +import org.apache.hadoop.hive.serde2.objectinspector.StructObjectInspector;
    +import com.google.common.collect.Lists;
    +import java.util.ArrayList;
    +import java.util.Arrays;
    +import java.util.HashSet;
    +import java.util.Queue;
    +import java.util.Set;
    +import org.apache.hadoop.hive.metastore.api.hive_metastoreConstants;
    +import org.apache.hadoop.hive.serde.serdeConstants;
    +</#if>
    +
    +public class Hive${entry.hiveReader}Reader extends HiveAbstractReader {
    +
    +  Object key;
    +<#if entry.hasHeaderFooter == true>
    --- End diff --
    
    Free marker allows simple boolean expressions:
    `<#if entry.hasHeaderFooter>`


---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at infrastructure@apache.org or file a JIRA ticket
with INFRA.
---

[GitHub] drill pull request #638: DRILL-4982: Separate Hive reader classes for differ...

Posted by paul-rogers <gi...@git.apache.org>.
Github user paul-rogers commented on a diff in the pull request:

    https://github.com/apache/drill/pull/638#discussion_r91447926
  
    --- Diff: contrib/storage-hive/core/src/main/codegen/templates/HiveRecordReaders.java ---
    @@ -0,0 +1,300 @@
    +/**
    + * Licensed to the Apache Software Foundation (ASF) under one
    + * or more contributor license agreements.  See the NOTICE file
    + * distributed with this work for additional information
    + * regarding copyright ownership.  The ASF licenses this file
    + * to you under the Apache License, Version 2.0 (the
    + * "License"); you may not use this file except in compliance
    + * with the License.  You may obtain a copy of the License at
    + *
    + * http://www.apache.org/licenses/LICENSE-2.0
    + *
    + * Unless required by applicable law or agreed to in writing, software
    + * distributed under the License is distributed on an "AS IS" BASIS,
    + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
    + * See the License for the specific language governing permissions and
    + * limitations under the License.
    + */
    +
    +/**
    + * This template is used to generate different Hive record reader classes for different data formats
    + * to avoid JIT profile pullusion. These readers are derived from HiveAbstractReader which implements
    + * codes for init and setup stage, but the repeated - and performance critical part - next() method is
    + * separately implemented in the classes generated from this template. The internal SkipRecordReeader
    + * class is also separated as well due to the same reason.
    + *
    + * As to the performance gain with this change, please refer to:
    + * https://issues.apache.org/jira/browse/DRILL-4982
    + *
    + */
    +<@pp.dropOutputFile />
    +<#list hiveFormat.map as entry>
    +<@pp.changeOutputFile name="/org/apache/drill/exec/store/hive/Hive${entry.hiveReader}Reader.java" />
    +<#include "/@includes/license.ftl" />
    +
    +package org.apache.drill.exec.store.hive;
    +
    +import java.io.IOException;
    +import java.util.List;
    +import java.util.Properties;
    +import org.apache.drill.common.exceptions.DrillRuntimeException;
    +import org.apache.drill.common.exceptions.ExecutionSetupException;
    +import org.apache.drill.common.expression.SchemaPath;
    +import org.apache.drill.exec.ops.FragmentContext;
    +import org.apache.drill.exec.vector.AllocationHelper;
    +import org.apache.drill.exec.vector.ValueVector;
    +import org.apache.hadoop.hive.metastore.api.Partition;
    +import org.apache.hadoop.hive.metastore.api.Table;
    +import org.apache.hadoop.io.Writable;
    +import org.apache.hadoop.mapred.InputSplit;
    +import org.apache.hadoop.security.UserGroupInformation;
    +import org.apache.hadoop.hive.conf.HiveConf;
    +
    +import org.apache.hadoop.hive.serde2.SerDeException;
    +
    +import org.apache.hadoop.mapred.RecordReader;
    +<#if entry.hasHeaderFooter == true>
    +import org.apache.hadoop.hive.serde2.SerDe;
    +import org.apache.hadoop.hive.serde2.objectinspector.StructObjectInspector;
    +import com.google.common.collect.Lists;
    +import java.util.ArrayList;
    +import java.util.Arrays;
    +import java.util.HashSet;
    +import java.util.Queue;
    +import java.util.Set;
    +import org.apache.hadoop.hive.metastore.api.hive_metastoreConstants;
    +import org.apache.hadoop.hive.serde.serdeConstants;
    +</#if>
    +
    +public class Hive${entry.hiveReader}Reader extends HiveAbstractReader {
    +
    +  Object key;
    +<#if entry.hasHeaderFooter == true>
    +  SkipRecordsInspector skipRecordsInspector;
    +<#else>
    +  Object value;
    +</#if>
    +
    +  public Hive${entry.hiveReader}Reader(Table table, Partition partition, InputSplit inputSplit, List<SchemaPath> projectedColumns,
    +                       FragmentContext context, final HiveConf hiveConf,
    +                       UserGroupInformation proxyUgi) throws ExecutionSetupException {
    +    super(table, partition, inputSplit, projectedColumns, context, hiveConf, proxyUgi);
    +  }
    +
    +  public  void internalInit(Properties tableProperties, RecordReader<Object, Object> reader) {
    +
    +    key = reader.createKey();
    +<#if entry.hasHeaderFooter == true>
    +    skipRecordsInspector = new SkipRecordsInspector(tableProperties, reader);
    +<#else>
    +    value = reader.createValue();
    +</#if>
    +
    +  }
    +  private void readHiveRecordAndInsertIntoRecordBatch(Object deSerializedValue, int outputRecordIndex) {
    +    for (int i = 0; i < selectedStructFieldRefs.size(); i++) {
    +      Object hiveValue = finalOI.getStructFieldData(deSerializedValue, selectedStructFieldRefs.get(i));
    +      if (hiveValue != null) {
    +        selectedColumnFieldConverters.get(i).setSafeValue(selectedColumnObjInspectors.get(i), hiveValue,
    +          vectors.get(i), outputRecordIndex);
    +      }
    +    }
    +  }
    +
    +<#if entry.hasHeaderFooter == true>
    +  @Override
    +  public int next() {
    +    for (ValueVector vv : vectors) {
    +      AllocationHelper.allocateNew(vv, TARGET_RECORD_COUNT);
    +    }
    +    if (empty) {
    +      setValueCountAndPopulatePartitionVectors(0);
    +      return 0;
    +    }
    +
    +    try {
    +      skipRecordsInspector.reset();
    +      Object value;
    +
    +      int recordCount = 0;
    +
    +      while (recordCount < TARGET_RECORD_COUNT && reader.next(key, value = skipRecordsInspector.getNextValue())) {
    +        if (skipRecordsInspector.doSkipHeader(recordCount++)) {
    +          continue;
    +        }
    +        Object bufferedValue = skipRecordsInspector.bufferAdd(value);
    +        if (bufferedValue != null) {
    +          Object deSerializedValue = partitionSerDe.deserialize((Writable) bufferedValue);
    +          if (partTblObjectInspectorConverter != null) {
    +            deSerializedValue = partTblObjectInspectorConverter.convert(deSerializedValue);
    +          }
    +          readHiveRecordAndInsertIntoRecordBatch(deSerializedValue, skipRecordsInspector.getActualCount());
    +          skipRecordsInspector.incrementActualCount();
    +        }
    +        skipRecordsInspector.incrementTempCount();
    +      }
    +
    +      setValueCountAndPopulatePartitionVectors(skipRecordsInspector.getActualCount());
    +      skipRecordsInspector.updateContinuance();
    +      return skipRecordsInspector.getActualCount();
    +    } catch (IOException | SerDeException e) {
    +      throw new DrillRuntimeException(e);
    +    }
    +  }
    +
    +/**
    + * SkipRecordsInspector encapsulates logic to skip header and footer from file.
    + * Logic is applicable only for predefined in constructor file formats.
    + */
    +protected class SkipRecordsInspector {
    --- End diff --
    
    Is there anything different in this nested class across the generated classes? If not, Can you explain why we want multiple copies of the same code? A quick comment will help greatly.


---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at infrastructure@apache.org or file a JIRA ticket
with INFRA.
---

[GitHub] drill pull request #638: DRILL-4982: Separate Hive reader classes for differ...

Posted by paul-rogers <gi...@git.apache.org>.
Github user paul-rogers commented on a diff in the pull request:

    https://github.com/apache/drill/pull/638#discussion_r91447651
  
    --- Diff: contrib/storage-hive/core/src/main/codegen/templates/HiveRecordReaders.java ---
    @@ -0,0 +1,300 @@
    +/**
    + * Licensed to the Apache Software Foundation (ASF) under one
    + * or more contributor license agreements.  See the NOTICE file
    + * distributed with this work for additional information
    + * regarding copyright ownership.  The ASF licenses this file
    + * to you under the Apache License, Version 2.0 (the
    + * "License"); you may not use this file except in compliance
    + * with the License.  You may obtain a copy of the License at
    + *
    + * http://www.apache.org/licenses/LICENSE-2.0
    + *
    + * Unless required by applicable law or agreed to in writing, software
    + * distributed under the License is distributed on an "AS IS" BASIS,
    + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
    + * See the License for the specific language governing permissions and
    + * limitations under the License.
    + */
    +
    +/**
    + * This template is used to generate different Hive record reader classes for different data formats
    + * to avoid JIT profile pullusion. These readers are derived from HiveAbstractReader which implements
    + * codes for init and setup stage, but the repeated - and performance critical part - next() method is
    + * separately implemented in the classes generated from this template. The internal SkipRecordReeader
    + * class is also separated as well due to the same reason.
    + *
    + * As to the performance gain with this change, please refer to:
    + * https://issues.apache.org/jira/browse/DRILL-4982
    + *
    + */
    +<@pp.dropOutputFile />
    +<#list hiveFormat.map as entry>
    +<@pp.changeOutputFile name="/org/apache/drill/exec/store/hive/Hive${entry.hiveReader}Reader.java" />
    +<#include "/@includes/license.ftl" />
    +
    +package org.apache.drill.exec.store.hive;
    +
    +import java.io.IOException;
    +import java.util.List;
    +import java.util.Properties;
    +import org.apache.drill.common.exceptions.DrillRuntimeException;
    +import org.apache.drill.common.exceptions.ExecutionSetupException;
    +import org.apache.drill.common.expression.SchemaPath;
    +import org.apache.drill.exec.ops.FragmentContext;
    +import org.apache.drill.exec.vector.AllocationHelper;
    +import org.apache.drill.exec.vector.ValueVector;
    +import org.apache.hadoop.hive.metastore.api.Partition;
    +import org.apache.hadoop.hive.metastore.api.Table;
    +import org.apache.hadoop.io.Writable;
    +import org.apache.hadoop.mapred.InputSplit;
    +import org.apache.hadoop.security.UserGroupInformation;
    +import org.apache.hadoop.hive.conf.HiveConf;
    +
    +import org.apache.hadoop.hive.serde2.SerDeException;
    +
    +import org.apache.hadoop.mapred.RecordReader;
    +<#if entry.hasHeaderFooter == true>
    +import org.apache.hadoop.hive.serde2.SerDe;
    +import org.apache.hadoop.hive.serde2.objectinspector.StructObjectInspector;
    +import com.google.common.collect.Lists;
    +import java.util.ArrayList;
    +import java.util.Arrays;
    +import java.util.HashSet;
    +import java.util.Queue;
    +import java.util.Set;
    +import org.apache.hadoop.hive.metastore.api.hive_metastoreConstants;
    +import org.apache.hadoop.hive.serde.serdeConstants;
    +</#if>
    +
    +public class Hive${entry.hiveReader}Reader extends HiveAbstractReader {
    +
    +  Object key;
    +<#if entry.hasHeaderFooter == true>
    +  SkipRecordsInspector skipRecordsInspector;
    +<#else>
    +  Object value;
    +</#if>
    +
    +  public Hive${entry.hiveReader}Reader(Table table, Partition partition, InputSplit inputSplit, List<SchemaPath> projectedColumns,
    +                       FragmentContext context, final HiveConf hiveConf,
    +                       UserGroupInformation proxyUgi) throws ExecutionSetupException {
    +    super(table, partition, inputSplit, projectedColumns, context, hiveConf, proxyUgi);
    +  }
    +
    +  public  void internalInit(Properties tableProperties, RecordReader<Object, Object> reader) {
    +
    +    key = reader.createKey();
    +<#if entry.hasHeaderFooter == true>
    +    skipRecordsInspector = new SkipRecordsInspector(tableProperties, reader);
    +<#else>
    +    value = reader.createValue();
    +</#if>
    +
    +  }
    +  private void readHiveRecordAndInsertIntoRecordBatch(Object deSerializedValue, int outputRecordIndex) {
    +    for (int i = 0; i < selectedStructFieldRefs.size(); i++) {
    +      Object hiveValue = finalOI.getStructFieldData(deSerializedValue, selectedStructFieldRefs.get(i));
    +      if (hiveValue != null) {
    +        selectedColumnFieldConverters.get(i).setSafeValue(selectedColumnObjInspectors.get(i), hiveValue,
    +          vectors.get(i), outputRecordIndex);
    +      }
    +    }
    +  }
    +
    +<#if entry.hasHeaderFooter == true>
    +  @Override
    +  public int next() {
    +    for (ValueVector vv : vectors) {
    +      AllocationHelper.allocateNew(vv, TARGET_RECORD_COUNT);
    +    }
    +    if (empty) {
    +      setValueCountAndPopulatePartitionVectors(0);
    +      return 0;
    +    }
    +
    +    try {
    +      skipRecordsInspector.reset();
    +      Object value;
    +
    +      int recordCount = 0;
    +
    +      while (recordCount < TARGET_RECORD_COUNT && reader.next(key, value = skipRecordsInspector.getNextValue())) {
    +        if (skipRecordsInspector.doSkipHeader(recordCount++)) {
    +          continue;
    +        }
    +        Object bufferedValue = skipRecordsInspector.bufferAdd(value);
    +        if (bufferedValue != null) {
    +          Object deSerializedValue = partitionSerDe.deserialize((Writable) bufferedValue);
    +          if (partTblObjectInspectorConverter != null) {
    +            deSerializedValue = partTblObjectInspectorConverter.convert(deSerializedValue);
    +          }
    +          readHiveRecordAndInsertIntoRecordBatch(deSerializedValue, skipRecordsInspector.getActualCount());
    +          skipRecordsInspector.incrementActualCount();
    +        }
    +        skipRecordsInspector.incrementTempCount();
    +      }
    +
    +      setValueCountAndPopulatePartitionVectors(skipRecordsInspector.getActualCount());
    +      skipRecordsInspector.updateContinuance();
    +      return skipRecordsInspector.getActualCount();
    +    } catch (IOException | SerDeException e) {
    +      throw new DrillRuntimeException(e);
    +    }
    +  }
    +
    +/**
    + * SkipRecordsInspector encapsulates logic to skip header and footer from file.
    + * Logic is applicable only for predefined in constructor file formats.
    + */
    +protected class SkipRecordsInspector {
    --- End diff --
    
    Small point, but nested classes are usually indented one level to help highlight that they are, in fact, nested.


---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at infrastructure@apache.org or file a JIRA ticket
with INFRA.
---

[GitHub] drill pull request #638: DRILL-4982: Separate Hive reader classes for differ...

Posted by paul-rogers <gi...@git.apache.org>.
Github user paul-rogers commented on a diff in the pull request:

    https://github.com/apache/drill/pull/638#discussion_r91448234
  
    --- Diff: contrib/storage-hive/core/src/main/codegen/data/HiveFormats.tdd ---
    @@ -0,0 +1,50 @@
    +# Licensed to the Apache Software Foundation (ASF) under one
    +# or more contributor license agreements.  See the NOTICE file
    +# distributed with this work for additional information
    +# regarding copyright ownership.  The ASF licenses this file
    +# to you under the Apache License, Version 2.0 (the
    +# "License"); you may not use this file except in compliance
    +# with the License.  You may obtain a copy of the License at
    +#
    +# http:# www.apache.org/licenses/LICENSE-2.0
    +#
    +# Unless required by applicable law or agreed to in writing, software
    +# distributed under the License is distributed on an "AS IS" BASIS,
    +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
    +# See the License for the specific language governing permissions and
    +# limitations under the License.
    +
    +{
    --- End diff --
    
    Can we explain this a bit? We have 6 reader types. But, the only difference in generated code is has header/footer or not. Can we solve the Java optimization problem with a "classic" type hierarchy:
    
    ```
    HiveAbstractReader
    . HiveSimpleReader
    . . HiveAvroReader
    . . ...
    . HiveHeaderFooterReader
    . . HiveTextReader
    . . ...
    ```
    
    Names are just made up. The point is, can a much simpler Java hierarchy, with less duplicated code, solve the problem? If there is one function that is sub-optimized, can just that one function be generated in the subclass rather than generating duplicate code?


---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at infrastructure@apache.org or file a JIRA ticket
with INFRA.
---