store query command list in task (#3649)

Sometimes we have concatenated query strings for a task. However,
when we want to find each query string, it is not a trivial task.
Therefore, it makes sense to store this in task so that when we need
each query string we can easily get it.

src/backend/distributed/commands/utility_hook.c

@@ -927,7 +927,6 @@ List *
 NodeDDLTaskList(TargetWorkerSet targets, List *commands)
 {
 	List *workerNodes = TargetWorkerSetNodeList(targets, NoLock);
-	char *concatenatedCommands = StringJoin(commands, ';');
 
 	if (list_length(workerNodes) <= 0)
 	{
@@ -940,7 +939,7 @@ NodeDDLTaskList(TargetWorkerSet targets, List *commands)
 
 	Task *task = CitusMakeNode(Task);
 	task->taskType = DDL_TASK;
-	SetTaskQueryString(task, concatenatedCommands);
+	SetTaskQueryStringList(task, commands);
 
 	WorkerNode *workerNode = NULL;
 	foreach_ptr(workerNode, workerNodes)

src/backend/distributed/master/master_stage_protocol.c

@@ -513,7 +513,8 @@ CreateShardsOnWorkers(Oid distributedRelationId, List *shardPlacements,
 		task->jobId = INVALID_JOB_ID;
 		task->taskId = taskId++;
 		task->taskType = DDL_TASK;
-		SetTaskQueryString(task, StringJoin(commandList, ';'));
+		SetTaskQueryStringList(task, commandList);
+
 		task->replicationModel = REPLICATION_MODEL_INVALID;
 		task->dependentTaskList = NIL;
 		task->anchorShardId = shardId;

src/backend/distributed/planner/deparse_shard_query.c

@@ -445,6 +445,17 @@ SetTaskQueryString(Task *task, char *queryString)
 }
 
 
+/*
+ * SetTaskQueryStringList sets the queryStringList of the given task.
+ */
+void
+SetTaskQueryStringList(Task *task, List *queryStringList)
+{
+	task->queryStringList = queryStringList;
+	SetTaskQueryString(task, StringJoin(queryStringList, ';'));
+}
+
+
 /*
  * DeparseTaskQuery is a general way of deparsing a query based on a task.
  */

src/backend/distributed/utils/citus_copyfuncs.c

@@ -260,6 +260,7 @@ CopyNodeTask(COPYFUNC_ARGS)
 	COPY_SCALAR_FIELD(taskId);
 	COPY_NODE_FIELD(queryForLocalExecution);
 	COPY_STRING_FIELD(queryStringLazy);
+	COPY_NODE_FIELD(queryStringList);
 	COPY_SCALAR_FIELD(anchorDistributedTableId);
 	COPY_SCALAR_FIELD(anchorShardId);
 	COPY_NODE_FIELD(taskPlacementList);

src/backend/distributed/utils/citus_outfuncs.c

@@ -481,6 +481,7 @@ OutTask(OUTFUNC_ARGS)
 	WRITE_UINT_FIELD(taskId);
 	WRITE_NODE_FIELD(queryForLocalExecution);
 	WRITE_STRING_FIELD(queryStringLazy);
+	WRITE_NODE_FIELD(queryStringList);
 	WRITE_OID_FIELD(anchorDistributedTableId);
 	WRITE_UINT64_FIELD(anchorShardId);
 	WRITE_NODE_FIELD(taskPlacementList);

src/include/distributed/deparse_shard_query.h

@@ -25,6 +25,7 @@ extern void RebuildQueryStrings(Job *workerJob);
 extern bool UpdateRelationToShardNames(Node *node, List *relationShardList);
 extern void SetTaskQuery(Task *task, Query *query);
 extern void SetTaskQueryString(Task *task, char *queryString);
+extern void SetTaskQueryStringList(Task *task, List *queryStringList);
 extern char * TaskQueryString(Task *task);
 extern bool UpdateRelationsToLocalShardTables(Node *node, List *relationShardList);
 

src/include/distributed/multi_physical_planner.h

@@ -240,6 +240,14 @@ typedef struct Task
 	char *queryStringLazy;
 	List *perPlacementQueryStrings;
 
+	/*
+	 * queryStringList contains query strings. They should be
+	 * run sequentially. The concatenated version of this list
+	 * will already be set for queryStringLazy, this can be useful
+	 * when we want to access each query string.
+	 */
+	List *queryStringList;
+
 	Oid anchorDistributedTableId;     /* only applies to insert tasks */
 	uint64 anchorShardId;       /* only applies to compute tasks */
 	List *taskPlacementList;    /* only applies to compute tasks */