paul-rogers commented on a change in pull request #11828:
URL: https://github.com/apache/druid/pull/11828#discussion_r747821547



##########
File path: 
processing/src/main/java/org/apache/druid/query/context/ResponseContext.java
##########
@@ -30,93 +33,334 @@
 import org.apache.druid.java.util.common.IAE;
 import org.apache.druid.java.util.common.ISE;
 import org.apache.druid.java.util.common.NonnullPair;
-import org.apache.druid.java.util.common.jackson.JacksonUtils;
 import org.apache.druid.query.SegmentDescriptor;
+import org.apache.druid.utils.CollectionUtils;
 import org.joda.time.Interval;
 
 import javax.annotation.Nullable;
+
 import java.io.IOException;
 import java.util.ArrayList;
-import java.util.Collection;
-import java.util.Collections;
 import java.util.Comparator;
 import java.util.List;
 import java.util.Map;
 import java.util.concurrent.ConcurrentHashMap;
 import java.util.concurrent.ConcurrentMap;
 import java.util.concurrent.ConcurrentSkipListMap;
-import java.util.function.BiFunction;
+import java.util.concurrent.atomic.AtomicLong;
+import java.util.function.Function;
+import java.util.stream.Collectors;
 
 /**
  * The context for storing and passing data between chains of {@link 
org.apache.druid.query.QueryRunner}s.
  * The context is also transferred between Druid nodes with all the data it 
contains.
- */
+ * <p>
+ * The response context consists of a set of key/value pairs. Keys are those 
defined in
+ * the {@code Keys} registry. Keys are indexed by key instance, not by name. 
The
+ * key defines the type of the associated value, including logic to merge 
values and
+ * to deserialize JSON values for that key.
+ *
+ * <h4>Structure</h4>
+ * The context has evolved to perform multiple tasks. First, it holds two kinds
+ * of information:
+ * <ul>
+ * <li>Information to be returned in the query response header.
+ * (These are values tagged as {@code HEADER}.)</li>
+ * <li>Values passed within a single server. These are tagged with
+ * visibility {@code NONE}.)</li>
+ * </ul>
+ * Second, it performs multiple tasks:
+ * <ul>
+ * <li>Registers the keys to be used in the header. But, since it also holds
+ * internal information, the internal information also needs keys, though the
+ * corresponding values are never serialized.</li>
+ * <li>Gathers information for the query as a whole.</li>
+ * <li>Merges information back up the query tree: from multiple segments,
+ * from multiple servers, etc.</li>
+ * <li>Manages headers size by dropping fields when the header would get too
+ * large.</li>
+ * </ul>
+ *
+ * A result is that the information the context, when inspected by a calling
+ * query, may be incomplete if some of it was previously dropped by the
+ * called query.
+ *
+ * <h4>API</h4>
+ *
+ * The query profile needs to obtain the full, untruncated information. To do 
this
+ * it piggy-backs on the set operations to obtain the full value. To ensure 
this
+ * is possible, code that works with standard values should call the set (or 
add)
+ * functions provided which will do the needed map update.
+  */
 @PublicApi
 public abstract class ResponseContext
 {
   /**
    * The base interface of a response context key.
    * Should be implemented by every context key.
    */
-  public interface BaseKey
+  public interface Key

Review comment:
       @clintropolis, you are right. As it turned out, none of the extensions 
in the Druid code repo itself actually had this issue, which suggested that, 
while this is an extension point, it may not actually have many (or any) 
extensions. So, the benefit of the simpler code (especially as we add the 
response trailer) seemed worth the risk.
   
   I wonder, how do we normally communicate such potentially breaking changes? 




-- 
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.

To unsubscribe, e-mail: [email protected]

For queries about this service, please contact Infrastructure at:
[email protected]



---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]

Reply via email to