diff --git a/docs/0.9.5/sqlpp/manual.html b/docs/0.9.5/sqlpp/manual.html
new file mode 100644
index 0000000..8d62f44
--- /dev/null
+++ b/docs/0.9.5/sqlpp/manual.html
@@ -0,0 +1,4935 @@
+<!DOCTYPE html>
+<!--
+ | Generated by Apache Maven Doxia Site Renderer 1.8.1 from target/generated-site/markdown/sqlpp/manual.md at 2020-08-07
+ | Rendered using Apache Maven Fluido Skin 1.7
+-->
+<html xmlns="http://www.w3.org/1999/xhtml" xml:lang="en" lang="en">
+  <head>
+    <meta charset="UTF-8" />
+    <meta name="viewport" content="width=device-width, initial-scale=1.0" />
+    <meta name="Date-Revision-yyyymmdd" content="20200807" />
+    <meta http-equiv="Content-Language" content="en" />
+    <title>AsterixDB &#x2013; The Query Language</title>
+    <link rel="stylesheet" href="../css/apache-maven-fluido-1.7.min.css" />
+    <link rel="stylesheet" href="../css/site.css" />
+    <link rel="stylesheet" href="../css/print.css" media="print" />
+    <script type="text/javascript" src="../js/apache-maven-fluido-1.7.min.js"></script>
+
+  </head>
+  <body class="topBarDisabled">
+    <div class="container-fluid">
+      <div id="banner">
+        <div class="pull-left"><a href=".././" id="bannerLeft"><img src="../images/asterixlogo.png"  alt="AsterixDB"/></a></div>
+        <div class="pull-right"></div>
+        <div class="clear"><hr/></div>
+      </div>
+
+      <div id="breadcrumbs">
+        <ul class="breadcrumb">
+        <li id="publishDate">Last Published: 2020-08-07</li>
+      <li id="projectVersion" class="pull-right">Version: 0.9.5</li>
+      <li class="pull-right"><a href="../index.html" title="Documentation Home">Documentation Home</a></li>
+        </ul>
+      </div>
+      <div class="row-fluid">
+        <div id="leftColumn" class="span2">
+          <div class="well sidebar-nav">
+    <ul class="nav nav-list">
+      <li class="nav-header">Get Started - Installation</li>
+    <li><a href="../ncservice.html" title="Option 1: using NCService"><span class="none"></span>Option 1: using NCService</a></li>
+    <li><a href="../ansible.html" title="Option 2: using Ansible"><span class="none"></span>Option 2: using Ansible</a></li>
+    <li><a href="../aws.html" title="Option 3: using Amazon Web Services"><span class="none"></span>Option 3: using Amazon Web Services</a></li>
+      <li class="nav-header">AsterixDB Primer</li>
+    <li><a href="../sqlpp/primer-sqlpp.html" title="Using SQL++"><span class="none"></span>Using SQL++</a></li>
+      <li class="nav-header">Data Model</li>
+    <li><a href="../datamodel.html" title="The Asterix Data Model"><span class="none"></span>The Asterix Data Model</a></li>
+      <li class="nav-header">Queries</li>
+    <li class="active"><a href="#"><span class="none"></span>The SQL++ Query Language</a></li>
+    <li><a href="../sqlpp/builtins.html" title="Builtin Functions"><span class="none"></span>Builtin Functions</a></li>
+      <li class="nav-header">API/SDK</li>
+    <li><a href="../api.html" title="HTTP API"><span class="none"></span>HTTP API</a></li>
+    <li><a href="../csv.html" title="CSV Output"><span class="none"></span>CSV Output</a></li>
+      <li class="nav-header">Advanced Features</li>
+    <li><a href="../aql/externaldata.html" title="Accessing External Data"><span class="none"></span>Accessing External Data</a></li>
+    <li><a href="../feeds.html" title="Data Ingestion with Feeds"><span class="none"></span>Data Ingestion with Feeds</a></li>
+    <li><a href="../udf.html" title="User Defined Functions"><span class="none"></span>User Defined Functions</a></li>
+    <li><a href="../sqlpp/filters.html" title="Filter-Based LSM Index Acceleration"><span class="none"></span>Filter-Based LSM Index Acceleration</a></li>
+    <li><a href="../sqlpp/fulltext.html" title="Support of Full-text Queries"><span class="none"></span>Support of Full-text Queries</a></li>
+    <li><a href="../sqlpp/similarity.html" title="Support of Similarity Queries"><span class="none"></span>Support of Similarity Queries</a></li>
+      <li class="nav-header">Deprecated</li>
+    <li><a href="../aql/primer.html" title="AsterixDB Primer: Using AQL"><span class="none"></span>AsterixDB Primer: Using AQL</a></li>
+    <li><a href="../aql/manual.html" title="Queries: The Asterix Query Language (AQL)"><span class="none"></span>Queries: The Asterix Query Language (AQL)</a></li>
+    <li><a href="../aql/builtins.html" title="Queries: Builtin Functions (AQL)"><span class="none"></span>Queries: Builtin Functions (AQL)</a></li>
+</ul>
+          <hr />
+          <div id="poweredBy">
+            <div class="clear"></div>
+            <div class="clear"></div>
+            <div class="clear"></div>
+            <div class="clear"></div>
+<a href=".././" title="AsterixDB" class="builtBy"><img class="builtBy"  alt="AsterixDB" src="../images/asterixlogo.png"    /></a>
+            </div>
+          </div>
+        </div>
+        <div id="bodyColumn"  class="span10" >
+<!--
+ ! 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.
+ !-->
+<h1>The Query Language</h1>
+<ul>
+
+<li><a href="#Introduction">1. Introduction</a></li>
+<li><a href="#Expressions">2. Expressions</a>
+<ul>
+
+<li><a href="#Operator_expressions">Operator Expressions</a>
+<ul>
+
+<li><a href="#Arithmetic_operators">Arithmetic Operators</a></li>
+<li><a href="#Collection_operators">Collection Operators</a></li>
+<li><a href="#Comparison_operators">Comparison Operators</a></li>
+<li><a href="#Logical_operators">Logical Operators</a></li>
+</ul>
+</li>
+<li><a href="#Quantified_expressions">Quantified Expressions</a></li>
+<li><a href="#Path_expressions">Path Expressions</a></li>
+<li><a href="#Primary_expressions">Primary Expressions</a>
+<ul>
+
+<li><a href="#Literals">Literals</a></li>
+<li><a href="#Variable_references">Variable References</a></li>
+<li><a href="#Parenthesized_expressions">Parenthesized Expressions</a></li>
+<li><a href="#Function_call_expressions">Function call Expressions</a></li>
+<li><a href="#Case_expressions">Case Expressions</a></li>
+<li><a href="#Constructors">Constructors</a></li>
+</ul>
+</li>
+</ul>
+</li>
+<li><a href="#Queries">3. Queries</a>
+<ul>
+
+<li><a href="#Declarations">Declarations</a></li>
+<li><a href="#SELECT_statements">SELECT Statements</a></li>
+<li><a href="#Select_clauses">SELECT Clauses</a>
+<ul>
+
+<li><a href="#Select_element">Select Element/Value/Raw</a></li>
+<li><a href="#SQL_select">SQL-style Select</a></li>
+<li><a href="#Select_star">Select *</a></li>
+<li><a href="#Select_distinct">Select Distinct</a></li>
+<li><a href="#Unnamed_projections">Unnamed Projections</a></li>
+<li><a href="#Abbreviated_field_access_expressions">Abbreviated Field Access Expressions</a></li>
+</ul>
+</li>
+<li><a href="#Unnest_clauses">UNNEST Clauses</a>
+<ul>
+
+<li><a href="#Inner_unnests">Inner Unnests</a></li>
+<li><a href="#Left_outer_unnests">Left Outer Unnests</a></li>
+<li><a href="#Expressing_joins_using_unnests">Expressing Joins Using Unnests</a></li>
+</ul>
+</li>
+<li><a href="#From_clauses">FROM clauses</a>
+<ul>
+
+<li><a href="#Binding_expressions">Binding Expressions</a></li>
+<li><a href="#Multiple_from_terms">Multiple From Terms</a></li>
+<li><a href="#Expressing_joins_using_from_terms">Expressing Joins Using From Terms</a></li>
+<li><a href="#Implicit_binding_variables">Implicit Binding Variables</a></li>
+</ul>
+</li>
+<li><a href="#Join_clauses">JOIN Clauses</a>
+<ul>
+
+<li><a href="#Inner_joins">Inner Joins</a></li>
+<li><a href="#Left_outer_joins">Left Outer Joins</a></li>
+</ul>
+</li>
+<li><a href="#Group_By_clauses">GROUP BY Clauses</a>
+<ul>
+
+<li><a href="#Group_variables">Group Variables</a></li>
+<li><a href="#Implicit_group_key_variables">Implicit Group Key Variables</a></li>
+<li><a href="#Implicit_group_variables">Implicit Group Variables</a></li>
+<li><a href="#Aggregation_functions">Aggregation Functions</a></li>
+<li><a href="#SQL-92_aggregation_functions">SQL-92 Aggregation Functions</a></li>
+<li><a href="#SQL-92_compliant_gby">SQL-92 Compliant GROUP BY Aggregations</a></li>
+<li><a href="#Column_aliases">Column Aliases</a></li>
+</ul>
+</li>
+<li><a href="#Where_having_clauses">WHERE Clauses and HAVING Clauses</a></li>
+<li><a href="#Order_By_clauses">ORDER BY Clauses</a></li>
+<li><a href="#Limit_clauses">LIMIT Clauses</a></li>
+<li><a href="#With_clauses">WITH Clauses</a></li>
+<li><a href="#Let_clauses">LET Clauses</a></li>
+<li><a href="#Union_all">UNION ALL</a></li>
+<li><a href="#Over_clauses">OVER Clauses</a>
+<ul>
+
+<li><a href="#Window_function_call">Window Function Call</a></li>
+<li><a href="#Window_function_options">Window Function Options</a></li>
+<li><a href="#Window_frame_variable">Window Frame Variable</a></li>
+<li><a href="#Window_definition">Window Definition</a></li>
+</ul>
+</li>
+<li><a href="#Vs_SQL-92">Differences from SQL-92</a></li>
+</ul>
+</li>
+<li><a href="#Errors">4. Errors</a>
+<ul>
+
+<li><a href="#Syntax_errors">Syntax Errors</a></li>
+<li><a href="#Identifier_resolution_errors">Identifier Resolution Errors</a></li>
+<li><a href="#Type_errors">Type Errors</a></li>
+<li><a href="#Resource_errors">Resource Errors</a></li>
+</ul>
+</li>
+<li><a href="#DDL_and_DML_statements">5. DDL and DML Statements</a>
+<ul>
+
+<li><a href="#Lifecycle_management_statements">Lifecycle Management Statements</a>
+<ul>
+
+<li><a href="#Dataverses">Dataverses</a></li>
+<li><a href="#Types">Types</a></li>
+<li><a href="#Datasets">Datasets</a></li>
+<li><a href="#Indices">Indices</a></li>
+<li><a href="#Functions">Functions</a></li>
+<li><a href="#Synonyms">Synonyms</a></li>
+<li><a href="#Removal">Removal</a></li>
+<li><a href="#Load_statement">Load Statement</a></li>
+</ul>
+</li>
+<li><a href="#Modification_statements">Modification Statements</a>
+<ul>
+
+<li><a href="#Inserts">Inserts</a></li>
+<li><a href="#Upserts">Upserts</a></li>
+<li><a href="#Deletes">Deletes</a></li>
+</ul>
+</li>
+</ul>
+</li>
+<li><a href="#Reserved_keywords">Appendix 1. Reserved Keywords</a></li>
+<li><a href="#Performance_tuning">Appendix 2. Performance Tuning</a>
+<ul>
+
+<li><a href="#Parallelism_parameter">Parallelism Parameter</a></li>
+<li><a href="#Memory_parameters">Memory Parameters</a></li>
+</ul>
+</li>
+<li><a href="#Variable_bindings_and_name_resolution">Appendix 3. Variable Bindings and Name Resolution</a></li>
+</ul><!--
+ ! 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.
+ !-->
+
+<h1><a name="Introduction" id="Introduction">1. Introduction</a><font size="3" /></h1>
+<p>This document is intended as a reference guide to the full syntax and semantics of AsterixDB&#x2019;s query language, a SQL-based language for working with semistructured data. The language is a derivative of SQL++, a declarative query language for JSON data which is largely backwards compatible with SQL. SQL++ originated from research in the FORWARD project at UC San Diego, and it has much in common with SQL; some differences exist due to the different data models that the two languages were designed to serve. SQL was designed for interacting with the flat, schema-ified world of relational databases, while SQL++ generalizes SQL to also handle nested data formats (like JSON) and the schema-optional (or even schema-less) data models of modern NoSQL and BigData systems.</p>
+<p>In the context of Apache AsterixDB, the query language is intended for working with the Asterix Data Model (<a href="../datamodel.html">ADM</a>), a data model based on a superset of JSON with an enriched and flexible type system. New AsterixDB users are encouraged to read and work through the (much friendlier) guide &#x201c;<a href="primer-sqlpp.html">AsterixDB 101: An ADM and SQL++ Primer</a>&#x201d; before attempting to make use of this document. In addition, readers are advised to read through the <a href="../datamodel.html">Asterix Data Model (ADM) reference guide</a> first as well, as an understanding of the data model is a prerequisite to understanding the query language.</p>
+<p>In what follows, we detail the features of the query language in a grammar-guided manner. We list and briefly explain each of the productions in the query grammar, offering examples (and results) for clarity.</p><!--
+ ! 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.
+ !-->
+
+<h1><a name="Expressions" id="Expressions">2. Expressions</a></h1><!--
+ ! 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.
+ !-->
+
+<p>The query language is a highly composable expression language. Each expression in the query language returns zero or more data model instances. There are three major kinds of expressions. At the topmost level, an expression can be an OperatorExpression (similar to a mathematical expression) or a QuantifiedExpression (which yields a boolean value). Each will be detailed as we explore the full grammar of the language.</p>
+
+<div>
+<div>
+<pre class="source">Expression ::= OperatorExpression | QuantifiedExpression
+</pre></div></div>
+
+<p>Note that in the following text, words enclosed in angle brackets denote keywords that are not case-sensitive.</p>
+<div class="section">
+<h2><a name="Operator_Expressions"></a><a name="Operator_expressions" id="Operator_expressions">Operator Expressions</a></h2>
+<p>Operators perform a specific operation on the input values or expressions. The syntax of an operator expression is as follows:</p>
+
+<div>
+<div>
+<pre class="source">OperatorExpression ::= PathExpression
+                       | Operator OperatorExpression
+                       | OperatorExpression Operator (OperatorExpression)?
+                       | OperatorExpression &lt;BETWEEN&gt; OperatorExpression &lt;AND&gt; OperatorExpression
+</pre></div></div>
+
+<p>The language provides a full set of operators that you can use within its statements. Here are the categories of operators:</p>
+<ul>
+
+<li><a href="#Arithmetic_operators">Arithmetic Operators</a>, to perform basic mathematical operations;</li>
+<li><a href="#Collection_operators">Collection Operators</a>, to evaluate expressions on collections or objects;</li>
+<li><a href="#Comparison_operators">Comparison Operators</a>, to compare two expressions;</li>
+<li><a href="#Logical_operators">Logical Operators</a>, to combine operators using Boolean logic.</li>
+</ul>
+<p>The following table summarizes the precedence order (from higher to lower) of the major unary and binary operators:</p>
+<table border="0" class="table table-striped">
+<thead>
+
+<tr class="a">
+<th> Operator                                                                    </th>
+<th> Operation </th></tr>
+</thead><tbody>
+
+<tr class="b">
+<td> EXISTS, NOT EXISTS                                                          </td>
+<td>  Collection emptiness testing </td></tr>
+<tr class="a">
+<td> ^                                                                           </td>
+<td>  Exponentiation  </td></tr>
+<tr class="b">
+<td> *, /, DIV, MOD (%)                                                          </td>
+<td>  Multiplication, division, modulo </td></tr>
+<tr class="a">
+<td> +, -                                                                        </td>
+<td>  Addition, subtraction  </td></tr>
+<tr class="b">
+<td> ||                                                                </td>
+<td>  String concatenation </td></tr>
+<tr class="a">
+<td> IS NULL, IS NOT NULL, IS MISSING, IS NOT MISSING, <br />IS UNKNOWN, IS NOT UNKNOWN, IS VALUED, IS NOT VALUED </td>
+<td> Unknown value comparison </td></tr>
+<tr class="b">
+<td> BETWEEN, NOT BETWEEN                                                        </td>
+<td> Range comparison (inclusive on both sides) </td></tr>
+<tr class="a">
+<td> =, !=, &lt;&gt;, &lt;, &gt;, &lt;=, &gt;=, LIKE, NOT LIKE, IN, NOT IN                             </td>
+<td> Comparison  </td></tr>
+<tr class="b">
+<td> NOT                                                                         </td>
+<td> Logical negation </td></tr>
+<tr class="a">
+<td> AND                                                                         </td>
+<td> Conjunction </td></tr>
+<tr class="b">
+<td> OR                                                                          </td>
+<td> Disjunction </td></tr>
+</tbody>
+</table>
+<p>In general, if any operand evaluates to a <tt>MISSING</tt> value, the enclosing operator will return <tt>MISSING</tt>; if none of operands evaluates to a <tt>MISSING</tt> value but there is an operand evaluates to a <tt>NULL</tt> value, the enclosing operator will return <tt>NULL</tt>. However, there are a few exceptions listed in <a href="#Comparison_operators">comparison operators</a> and <a href="#Logical_operators">logical operators</a>.</p>
+<div class="section">
+<h3><a name="Arithmetic_Operators"></a><a name="Arithmetic_operators" id="Arithmetic_operators">Arithmetic Operators</a></h3>
+<p>Arithmetic operators are used to exponentiate, add, subtract, multiply, and divide numeric values, or concatenate string values.</p>
+<table border="0" class="table table-striped">
+<thead>
+
+<tr class="a">
+<th> Operator     </th>
+<th>  Purpose                                                                </th>
+<th> Example    </th></tr>
+</thead><tbody>
+
+<tr class="b">
+<td> +, -         </td>
+<td>  As unary operators, they denote a <br />positive or negative expression </td>
+<td> SELECT VALUE -1; </td></tr>
+<tr class="a">
+<td> +, -         </td>
+<td>  As binary operators, they add or subtract                              </td>
+<td> SELECT VALUE 1 + 2; </td></tr>
+<tr class="b">
+<td> *            </td>
+<td>  Multiply                                                               </td>
+<td> SELECT VALUE 4 * 2; </td></tr>
+<tr class="a">
+<td> /            </td>
+<td>  Divide (returns a value of type <tt>double</tt> if both operands are integers)</td>
+<td> SELECT VALUE 5 / 2; </td></tr>
+<tr class="b">
+<td> DIV          </td>
+<td>  Divide (returns an integer value if both operands are integers)        </td>
+<td> SELECT VALUE 5 DIV 2; </td></tr>
+<tr class="a">
+<td> MOD (%)      </td>
+<td>  Modulo                                                                 </td>
+<td> SELECT VALUE 5 % 2; </td></tr>
+<tr class="b">
+<td> ^            </td>
+<td>  Exponentiation                                                         </td>
+<td> SELECT VALUE 2^3;       </td></tr>
+<tr class="a">
+<td> || </td>
+<td>  String concatenation                                                   </td>
+<td> SELECT VALUE &#x201c;ab&#x201d;||&#x201c;c&#x201d;||&#x201c;d&#x201d;;       </td></tr>
+</tbody>
+</table></div>
+<div class="section">
+<h3><a name="Collection_Operators"></a><a name="Collection_operators" id="Collection_operators">Collection Operators</a></h3>
+<p>Collection operators are used for membership tests (IN, NOT IN) or empty collection tests (EXISTS, NOT EXISTS).</p>
+<table border="0" class="table table-striped">
+<thead>
+
+<tr class="a">
+<th> Operator   </th>
+<th>  Purpose                                     </th>
+<th> Example    </th></tr>
+</thead><tbody>
+
+<tr class="b">
+<td> IN         </td>
+<td>  Membership test                             </td>
+<td> SELECT * FROM ChirpMessages cm <br />WHERE cm.user.lang IN [&#x201c;en&#x201d;, &#x201c;de&#x201d;]; </td></tr>
+<tr class="a">
+<td> NOT IN     </td>
+<td>  Non-membership test                         </td>
+<td> SELECT * FROM ChirpMessages cm <br />WHERE cm.user.lang NOT IN [&#x201c;en&#x201d;]; </td></tr>
+<tr class="b">
+<td> EXISTS     </td>
+<td>  Check whether a collection is not empty     </td>
+<td> SELECT * FROM ChirpMessages cm <br />WHERE EXISTS cm.referredTopics; </td></tr>
+<tr class="a">
+<td> NOT EXISTS </td>
+<td>  Check whether a collection is empty         </td>
+<td> SELECT * FROM ChirpMessages cm <br />WHERE NOT EXISTS cm.referredTopics; </td></tr>
+</tbody>
+</table></div>
+<div class="section">
+<h3><a name="Comparison_Operators"></a><a name="Comparison_operators" id="Comparison_operators">Comparison Operators</a></h3>
+<p>Comparison operators are used to compare values. The comparison operators fall into one of two sub-categories: missing value comparisons and regular value comparisons. The query language (and JSON) has two ways of representing missing information in a object - the presence of the field with a NULL for its value (as in SQL), and the absence of the field (which JSON permits). For example, the first of the following objects represents Jack, whose friend is Jill. In the other examples, Jake is friendless a la SQL, with a friend field that is NULL, while Joe is friendless in a more natural (for JSON) way, i.e., by not having a friend field.</p>
+<div class="section">
+<div class="section">
+<h5><a name="Examples"></a>Examples</h5>
+<p>{&#x201c;name&#x201d;: &#x201c;Jack&#x201d;, &#x201c;friend&#x201d;: &#x201c;Jill&#x201d;}</p>
+<p>{&#x201c;name&#x201d;: &#x201c;Jake&#x201d;, &#x201c;friend&#x201d;: NULL}</p>
+<p>{&#x201c;name&#x201d;: &#x201c;Joe&#x201d;}</p>
+<p>The following table enumerates all of the query language&#x2019;s comparison operators.</p>
+<table border="0" class="table table-striped">
+<thead>
+
+<tr class="a">
+<th> Operator       </th>
+<th>  Purpose                                       </th>
+<th> Example    </th></tr>
+</thead><tbody>
+
+<tr class="b">
+<td> IS NULL        </td>
+<td>  Test if a value is NULL                       </td>
+<td> SELECT * FROM ChirpMessages cm <br />WHERE cm.user.name IS NULL; </td></tr>
+<tr class="a">
+<td> IS NOT NULL    </td>
+<td>  Test if a value is not NULL                   </td>
+<td> SELECT * FROM ChirpMessages cm <br />WHERE cm.user.name IS NOT NULL; </td></tr>
+<tr class="b">
+<td> IS MISSING     </td>
+<td>  Test if a value is MISSING                    </td>
+<td> SELECT * FROM ChirpMessages cm <br />WHERE cm.user.name IS MISSING; </td></tr>
+<tr class="a">
+<td> IS NOT MISSING </td>
+<td>  Test if a value is not MISSING                </td>
+<td> SELECT * FROM ChirpMessages cm <br />WHERE cm.user.name IS NOT MISSING;</td></tr>
+<tr class="b">
+<td> IS UNKNOWN     </td>
+<td>  Test if a value is NULL or MISSING            </td>
+<td> SELECT * FROM ChirpMessages cm <br />WHERE cm.user.name IS UNKNOWN; </td></tr>
+<tr class="a">
+<td> IS NOT UNKNOWN </td>
+<td>  Test if a value is neither NULL nor MISSING   </td>
+<td> SELECT * FROM ChirpMessages cm <br />WHERE cm.user.name IS NOT UNKNOWN;</td></tr>
+<tr class="b">
+<td> IS KNOWN (IS VALUED) </td>
+<td>  Test if a value is neither NULL nor MISSING </td>
+<td> SELECT * FROM ChirpMessages cm <br />WHERE cm.user.name IS KNOWN; </td></tr>
+<tr class="a">
+<td> IS NOT KNOWN (IS NOT VALUED) </td>
+<td>  Test if a value is NULL or MISSING </td>
+<td> SELECT * FROM ChirpMessages cm <br />WHERE cm.user.name IS NOT KNOWN; </td></tr>
+<tr class="b">
+<td> BETWEEN        </td>
+<td>  Test if a value is between a start value and <br />a end value. The comparison is inclusive <br />to both start and end values. </td>
+<td>  SELECT * FROM ChirpMessages cm <br />WHERE cm.chirpId BETWEEN 10 AND 20;</td></tr>
+<tr class="a">
+<td> =              </td>
+<td>  Equality test                                 </td>
+<td> SELECT * FROM ChirpMessages cm <br />WHERE cm.chirpId=10; </td></tr>
+<tr class="b">
+<td> !=             </td>
+<td>  Inequality test                               </td>
+<td> SELECT * FROM ChirpMessages cm <br />WHERE cm.chirpId!=10;</td></tr>
+<tr class="a">
+<td> &lt;&gt;             </td>
+<td>  Inequality test                               </td>
+<td> SELECT * FROM ChirpMessages cm <br />WHERE cm.chirpId&lt;&gt;10;</td></tr>
+<tr class="b">
+<td> &lt;              </td>
+<td>  Less than                                     </td>
+<td> SELECT * FROM ChirpMessages cm <br />WHERE cm.chirpId&lt;10; </td></tr>
+<tr class="a">
+<td> &gt;              </td>
+<td>  Greater than                                  </td>
+<td> SELECT * FROM ChirpMessages cm <br />WHERE cm.chirpId&gt;10; </td></tr>
+<tr class="b">
+<td> &lt;=             </td>
+<td>  Less than or equal to                         </td>
+<td> SELECT * FROM ChirpMessages cm <br />WHERE cm.chirpId&lt;=10; </td></tr>
+<tr class="a">
+<td> &gt;=             </td>
+<td>  Greater than or equal to                      </td>
+<td> SELECT * FROM ChirpMessages cm <br />WHERE cm.chirpId&gt;=10; </td></tr>
+<tr class="b">
+<td> LIKE           </td>
+<td>  Test if the left side matches a<br /> pattern defined on the right<br /> side; in the pattern,  &#x201c;%&#x201d; matches  <br />any string while &#x201c;_&#x201d; matches <br /> any character. </td>
+<td> SELECT * FROM ChirpMessages cm <br />WHERE cm.user.name LIKE &#x201c;%Giesen%&#x201d;;</td></tr>
+<tr class="a">
+<td> NOT LIKE       </td>
+<td>  Test if the left side does not <br />match a pattern defined on the right<br /> side; in the pattern,  &#x201c;%&#x201d; matches <br />any string while &#x201c;_&#x201d; matches <br /> any character. </td>
+<td> SELECT * FROM ChirpMessages cm <br />WHERE cm.user.name NOT LIKE &#x201c;%Giesen%&#x201d;;</td></tr>
+</tbody>
+</table>
+<p>The following table summarizes how the missing value comparison operators work.</p>
+<table border="0" class="table table-striped">
+<thead>
+
+<tr class="a">
+<th> Operator </th>
+<th> Non-NULL/Non-MISSING value </th>
+<th> NULL </th>
+<th> MISSING </th></tr>
+</thead><tbody>
+
+<tr class="b">
+<td> IS NULL  </td>
+<td> FALSE </td>
+<td> TRUE </td>
+<td> MISSING </td></tr>
+<tr class="a">
+<td> IS NOT NULL </td>
+<td> TRUE </td>
+<td> FALSE </td>
+<td> MISSING </td></tr>
+<tr class="b">
+<td> IS MISSING  </td>
+<td> FALSE </td>
+<td> FALSE </td>
+<td> TRUE </td></tr>
+<tr class="a">
+<td> IS NOT MISSING </td>
+<td> TRUE </td>
+<td> TRUE </td>
+<td> FALSE </td></tr>
+<tr class="b">
+<td> IS UNKNOWN </td>
+<td> FALSE </td>
+<td> TRUE </td>
+<td> TRUE </td></tr>
+<tr class="a">
+<td> IS NOT UNKNOWN </td>
+<td> TRUE </td>
+<td> FALSE </td>
+<td> FALSE</td></tr>
+<tr class="b">
+<td> IS KNOWN (IS VALUED) </td>
+<td> TRUE </td>
+<td> FALSE </td>
+<td> FALSE </td></tr>
+<tr class="a">
+<td> IS NOT KNOWN (IS NOT VALUED) </td>
+<td> FALSE </td>
+<td> TRUE </td>
+<td> TRUE </td></tr>
+</tbody>
+</table></div></div></div>
+<div class="section">
+<h3><a name="Logical_Operators"></a><a name="Logical_operators" id="Logical_operators">Logical Operators</a></h3>
+<p>Logical operators perform logical <tt>NOT</tt>, <tt>AND</tt>, and <tt>OR</tt> operations over Boolean values (<tt>TRUE</tt> and <tt>FALSE</tt>) plus <tt>NULL</tt> and <tt>MISSING</tt>.</p>
+<table border="0" class="table table-striped">
+<thead>
+
+<tr class="a">
+<th> Operator </th>
+<th>  Purpose                                   </th>
+<th> Example    </th></tr>
+</thead><tbody>
+
+<tr class="b">
+<td> NOT      </td>
+<td>  Returns true if the following condition is false, otherwise returns false  </td>
+<td> SELECT VALUE NOT TRUE;  </td></tr>
+<tr class="a">
+<td> AND      </td>
+<td>  Returns true if both branches are true, otherwise returns false            </td>
+<td> SELECT VALUE TRUE AND FALSE; </td></tr>
+<tr class="b">
+<td> OR       </td>
+<td>  Returns true if one branch is true, otherwise returns false                </td>
+<td> SELECT VALUE FALSE OR FALSE; </td></tr>
+</tbody>
+</table>
+<p>The following table is the truth table for <tt>AND</tt> and <tt>OR</tt>.</p>
+<table border="0" class="table table-striped">
+<thead>
+
+<tr class="a">
+<th> A  </th>
+<th> B  </th>
+<th> A AND B  </th>
+<th> A OR B </th></tr>
+</thead><tbody>
+
+<tr class="b">
+<td> TRUE </td>
+<td> TRUE </td>
+<td> TRUE </td>
+<td> TRUE </td></tr>
+<tr class="a">
+<td> TRUE </td>
+<td> FALSE </td>
+<td> FALSE </td>
+<td> TRUE </td></tr>
+<tr class="b">
+<td> TRUE </td>
+<td> NULL </td>
+<td> NULL </td>
+<td> TRUE </td></tr>
+<tr class="a">
+<td> TRUE </td>
+<td> MISSING </td>
+<td> MISSING </td>
+<td> TRUE </td></tr>
+<tr class="b">
+<td> FALSE </td>
+<td> FALSE </td>
+<td> FALSE </td>
+<td> FALSE </td></tr>
+<tr class="a">
+<td> FALSE </td>
+<td> NULL </td>
+<td> FALSE </td>
+<td> NULL </td></tr>
+<tr class="b">
+<td> FALSE </td>
+<td> MISSING </td>
+<td> FALSE </td>
+<td> MISSING </td></tr>
+<tr class="a">
+<td> NULL </td>
+<td> NULL </td>
+<td> NULL </td>
+<td> NULL </td></tr>
+<tr class="b">
+<td> NULL </td>
+<td> MISSING </td>
+<td> MISSING </td>
+<td> NULL </td></tr>
+<tr class="a">
+<td> MISSING </td>
+<td> MISSING </td>
+<td> MISSING </td>
+<td> MISSING </td></tr>
+</tbody>
+</table>
+<p>The following table demonstrates the results of <tt>NOT</tt> on all possible inputs.</p>
+<table border="0" class="table table-striped">
+<thead>
+
+<tr class="a">
+<th> A  </th>
+<th> NOT A </th></tr>
+</thead><tbody>
+
+<tr class="b">
+<td> TRUE </td>
+<td> FALSE </td></tr>
+<tr class="a">
+<td> FALSE </td>
+<td> TRUE </td></tr>
+<tr class="b">
+<td> NULL </td>
+<td> NULL </td></tr>
+<tr class="a">
+<td> MISSING </td>
+<td> MISSING </td></tr>
+</tbody>
+</table></div></div>
+<div class="section">
+<h2><a name="Quantified_Expressions"></a><a name="Quantified_expressions" id="Quantified_expressions">Quantified Expressions</a></h2>
+
+<div>
+<div>
+<pre class="source">QuantifiedExpression ::= ( (&lt;ANY&gt;|&lt;SOME&gt;) | &lt;EVERY&gt; ) Variable &lt;IN&gt; Expression ( &quot;,&quot; Variable &quot;in&quot; Expression )*
+                         &lt;SATISFIES&gt; Expression (&lt;END&gt;)?
+</pre></div></div>
+
+<p>Quantified expressions are used for expressing existential or universal predicates involving the elements of a collection.</p>
+<p>The following pair of examples illustrate the use of a quantified expression to test that every (or some) element in the set [1, 2, 3] of integers is less than three. The first example yields <tt>FALSE</tt> and second example yields <tt>TRUE</tt>.</p>
+<p>It is useful to note that if the set were instead the empty set, the first expression would yield <tt>TRUE</tt> (&#x201c;every&#x201d; value in an empty set satisfies the condition) while the second expression would yield <tt>FALSE</tt> (since there isn&#x2019;t &#x201c;some&#x201d; value, as there are no values in the set, that satisfies the condition).</p>
+<p>A quantified expression will return a <tt>NULL</tt> (or <tt>MISSING</tt>) if the first expression in it evaluates to <tt>NULL</tt> (or <tt>MISSING</tt>). A type error will be raised if the first expression in a quantified expression does not return a collection.</p>
+<div class="section">
+<div class="section">
+<div class="section">
+<h5><a name="Examples"></a>Examples</h5>
+
+<div>
+<div>
+<pre class="source">EVERY x IN [ 1, 2, 3 ] SATISFIES x &lt; 3
+SOME x IN [ 1, 2, 3 ] SATISFIES x &lt; 3
+</pre></div></div>
+</div></div></div></div>
+<div class="section">
+<h2><a name="Path_Expressions"></a><a name="Path_expressions" id="Path_expressions">Path Expressions</a></h2>
+
+<div>
+<div>
+<pre class="source">PathExpression  ::= PrimaryExpression ( Field | Index )*
+Field           ::= &quot;.&quot; Identifier
+Index           ::= &quot;[&quot; Expression (&quot;:&quot; ( Expression )? )? &quot;]&quot;
+</pre></div></div>
+
+<p>Components of complex types in the data model are accessed via path expressions. Path access can be applied to the result of a query expression that yields an instance of a complex type, for example, an object or an array instance.</p>
+<p>For objects, path access is based on field names, and it accesses the field whose name was specified.<br /> For arrays, path access is based on (zero-based) array-style indexing. Array indexes can be used to retrieve either a single element from an array, or a whole subset of an array. Accessing a single element is achieved by providing a single index argument (zero-based element position), while obtaining a subset of an array is achieved by providing the <tt>start</tt> and <tt>end</tt> (zero-based) index positions; the returned subset is from position <tt>start</tt> to position <tt>end - 1</tt>; the <tt>end</tt> position argument is optional. If a position argument is negative then the element position is counted from the end of the array (<tt>-1</tt> addresses the last element, <tt>-2</tt> next to last, and so on). Multisets have similar behavior to arrays, except for retrieving arbitrary items as the order of items is not fixed in multisets.</p>
+<p>Attempts to access non-existent fields or out-of-bound array elements produce the special value <tt>MISSING</tt>. Type errors will be raised for inappropriate use of a path expression, such as applying a field accessor to a numeric value.</p>
+<p>The following examples illustrate field access for an object, index-based element access or subset retrieval of an array, and also a composition thereof.</p>
+<div class="section">
+<div class="section">
+<div class="section">
+<h5><a name="Examples"></a>Examples</h5>
+
+<div>
+<div>
+<pre class="source">({&quot;name&quot;: &quot;MyABCs&quot;, &quot;array&quot;: [ &quot;a&quot;, &quot;b&quot;, &quot;c&quot;]}).array
+
+([&quot;a&quot;, &quot;b&quot;, &quot;c&quot;])[2]
+
+([&quot;a&quot;, &quot;b&quot;, &quot;c&quot;])[-1]
+
+({&quot;name&quot;: &quot;MyABCs&quot;, &quot;array&quot;: [ &quot;a&quot;, &quot;b&quot;, &quot;c&quot;]}).array[2]
+
+([&quot;a&quot;, &quot;b&quot;, &quot;c&quot;])[0:2]
+
+([&quot;a&quot;, &quot;b&quot;, &quot;c&quot;])[0:]
+
+([&quot;a&quot;, &quot;b&quot;, &quot;c&quot;])[-2:-1]
+</pre></div></div>
+</div></div></div></div>
+<div class="section">
+<h2><a name="Primary_Expressions"></a><a name="Primary_expressions" id="Primary_expressions">Primary Expressions</a></h2>
+
+<div>
+<div>
+<pre class="source">PrimaryExpr ::= Literal
+              | VariableReference
+              | ParameterReference
+              | ParenthesizedExpression
+              | FunctionCallExpression
+              | CaseExpression
+              | Constructor
+</pre></div></div>
+
+<p>The most basic building block for any expression in the query language is PrimaryExpression. This can be a simple literal (constant) value, a reference to a query variable that is in scope, a parenthesized expression, a function call, or a newly constructed instance of the data model (such as a newly constructed object, array, or multiset of data model instances).</p></div>
+<div class="section">
+<h2><a name="Literals" id="Literals">Literals</a></h2>
+
+<div>
+<div>
+<pre class="source">Literal        ::= StringLiteral
+                   | IntegerLiteral
+                   | FloatLiteral
+                   | DoubleLiteral
+                   | &lt;NULL&gt;
+                   | &lt;MISSING&gt;
+                   | &lt;TRUE&gt;
+                   | &lt;FALSE&gt;
+StringLiteral  ::= &quot;\&quot;&quot; (
+                             &lt;EscapeQuot&gt;
+                           | &lt;EscapeBslash&gt;
+                           | &lt;EscapeSlash&gt;
+                           | &lt;EscapeBspace&gt;
+                           | &lt;EscapeFormf&gt;
+                           | &lt;EscapeNl&gt;
+                           | &lt;EscapeCr&gt;
+                           | &lt;EscapeTab&gt;
+                           | ~[&quot;\&quot;&quot;,&quot;\\&quot;])*
+                    &quot;\&quot;&quot;
+                    | &quot;\'&quot;(
+                             &lt;EscapeApos&gt;
+                           | &lt;EscapeBslash&gt;
+                           | &lt;EscapeSlash&gt;
+                           | &lt;EscapeBspace&gt;
+                           | &lt;EscapeFormf&gt;
+                           | &lt;EscapeNl&gt;
+                           | &lt;EscapeCr&gt;
+                           | &lt;EscapeTab&gt;
+                           | ~[&quot;\'&quot;,&quot;\\&quot;])*
+                      &quot;\'&quot;
+&lt;ESCAPE_Apos&gt;  ::= &quot;\\\'&quot;
+&lt;ESCAPE_Quot&gt;  ::= &quot;\\\&quot;&quot;
+&lt;EscapeBslash&gt; ::= &quot;\\\\&quot;
+&lt;EscapeSlash&gt;  ::= &quot;\\/&quot;
+&lt;EscapeBspace&gt; ::= &quot;\\b&quot;
+&lt;EscapeFormf&gt;  ::= &quot;\\f&quot;
+&lt;EscapeNl&gt;     ::= &quot;\\n&quot;
+&lt;EscapeCr&gt;     ::= &quot;\\r&quot;
+&lt;EscapeTab&gt;    ::= &quot;\\t&quot;
+
+IntegerLiteral ::= &lt;DIGITS&gt;
+&lt;DIGITS&gt;       ::= [&quot;0&quot; - &quot;9&quot;]+
+FloatLiteral   ::= &lt;DIGITS&gt; ( &quot;f&quot; | &quot;F&quot; )
+                 | &lt;DIGITS&gt; ( &quot;.&quot; &lt;DIGITS&gt; ( &quot;f&quot; | &quot;F&quot; ) )?
+                 | &quot;.&quot; &lt;DIGITS&gt; ( &quot;f&quot; | &quot;F&quot; )
+DoubleLiteral  ::= &lt;DIGITS&gt; &quot;.&quot; &lt;DIGITS&gt;
+                   | &quot;.&quot; &lt;DIGITS&gt;
+</pre></div></div>
+
+<p>Literals (constants) in a query can be strings, integers, floating point values, double values, boolean constants, or special constant values like <tt>NULL</tt> and <tt>MISSING</tt>. The <tt>NULL</tt> value is like a <tt>NULL</tt> in SQL; it is used to represent an unknown field value. The special value <tt>MISSING</tt> is only meaningful in the context of field accesses; it occurs when the accessed field simply does not exist at all in a object being accessed.</p>
+<p>The following are some simple examples of literals.</p>
+<div class="section">
+<div class="section">
+<div class="section">
+<h5><a name="Examples"></a>Examples</h5>
+
+<div>
+<div>
+<pre class="source">'a string'
+&quot;test string&quot;
+42
+</pre></div></div>
+
+<p>Different from standard SQL, double quotes play the same role as single quotes and may be used for string literals in queries as well.</p></div></div></div>
+<div class="section">
+<h3><a name="Variable_References"></a><a name="Variable_references" id="Variable_references">Variable References</a></h3>
+
+<div>
+<div>
+<pre class="source">VariableReference     ::= &lt;IDENTIFIER&gt; | &lt;DelimitedIdentifier&gt;
+&lt;IDENTIFIER&gt;          ::= (&lt;LETTER&gt; | &quot;_&quot;) (&lt;LETTER&gt; | &lt;DIGIT&gt; | &quot;_&quot; | &quot;$&quot;)*
+&lt;LETTER&gt;              ::= [&quot;A&quot; - &quot;Z&quot;, &quot;a&quot; - &quot;z&quot;]
+DelimitedIdentifier   ::= &quot;`&quot; (&lt;EscapeQuot&gt;
+                                | &lt;EscapeBslash&gt;
+                                | &lt;EscapeSlash&gt;
+                                | &lt;EscapeBspace&gt;
+                                | &lt;EscapeFormf&gt;
+                                | &lt;EscapeNl&gt;
+                                | &lt;EscapeCr&gt;
+                                | &lt;EscapeTab&gt;
+                                | ~[&quot;`&quot;,&quot;\\&quot;])*
+                          &quot;`&quot;
+</pre></div></div>
+
+<p>A variable in a query can be bound to any legal data model value. A variable reference refers to the value to which an in-scope variable is bound. (E.g., a variable binding may originate from one of the <tt>FROM</tt>, <tt>WITH</tt> or <tt>LET</tt> clauses of a <tt>SELECT</tt> statement or from an input parameter in the context of a function body.) Backticks, for example, `id`, are used for delimited identifiers. Delimiting is needed when a variable&#x2019;s desired name clashes with a keyword or includes characters not allowed in regular identifiers. More information on exactly how variable references are resolved can be found in the appendix section on Variable Resolution.</p>
+<div class="section">
+<div class="section">
+<h5><a name="Examples"></a>Examples</h5>
+
+<div>
+<div>
+<pre class="source">tweet
+id
+`SELECT`
+`my-function`
+</pre></div></div>
+</div></div></div>
+<div class="section">
+<h3><a name="Parameter_References"></a><a name="Parameter_references" id="Parameter_references">Parameter References</a></h3>
+
+<div>
+<div>
+<pre class="source">ParameterReference              ::= NamedParameterReference | PositionalParameterReference
+NamedParameterReference         ::= &quot;$&quot; (&lt;IDENTIFIER&gt; | &lt;DelimitedIdentifier&gt;)
+PositionalParameterReference    ::= (&quot;$&quot; &lt;DIGITS&gt;) | &quot;?&quot;
+</pre></div></div>
+
+<p>A statement parameter is an external variable which value is provided through the <a href="../api.html#queryservice">statement execution API</a>. An error will be raised if the parameter is not bound at the query execution time. Positional parameter numbering starts at 1. &#x201c;?&#x201d; parameters are interpreted as $1, .. $N in the order in which they appear in the statement.</p>
+<div class="section">
+<div class="section">
+<h5><a name="Examples"></a>Examples</h5>
+
+<div>
+<div>
+<pre class="source">$id
+$1
+?
+</pre></div></div>
+</div></div></div>
+<div class="section">
+<h3><a name="Parenthesized_Expressions"></a><a name="Parenthesized_expressions" id="Parenthesized_expressions">Parenthesized Expressions</a></h3>
+
+<div>
+<div>
+<pre class="source">ParenthesizedExpression ::= &quot;(&quot; Expression &quot;)&quot; | Subquery
+</pre></div></div>
+
+<p>An expression can be parenthesized to control the precedence order or otherwise clarify a query. For composability, a subquery is also an parenthesized expression.</p>
+<p>The following expression evaluates to the value 2.</p>
+<div class="section">
+<div class="section">
+<h5><a name="Example"></a>Example</h5>
+
+<div>
+<div>
+<pre class="source">( 1 + 1 )
+</pre></div></div>
+</div></div></div>
+<div class="section">
+<h3><a name="Function_Call_Expressions"></a><a name="Function_call_expressions" id="Function_call_expressions">Function Call Expressions</a></h3>
+
+<div>
+<div>
+<pre class="source">FunctionCallExpression ::= ( FunctionName &quot;(&quot; ( Expression ( &quot;,&quot; Expression )* )? &quot;)&quot; ) | WindowFunctionCall
+</pre></div></div>
+
+<p>Functions are included in the query language, like most languages, as a way to package useful functionality or to componentize complicated or reusable computations. A function call is a legal query expression that represents the value resulting from the evaluation of its body expression with the given parameter bindings; the parameter value bindings can themselves be any expressions in the query language.</p>
+<p>Note that Window functions, and aggregate functions used as window functions, have a more complex syntax. Window function calls are described in the section on <a href="#Over_clauses">OVER Clauses</a>.</p>
+<p>The following example is a (built-in) function call expression whose value is 8.</p>
+<div class="section">
+<div class="section">
+<h5><a name="Example"></a>Example</h5>
+
+<div>
+<div>
+<pre class="source">length('a string')
+</pre></div></div>
+</div></div></div></div>
+<div class="section">
+<h2><a name="Case_Expressions"></a><a name="Case_expressions" id="Case_expressions">Case Expressions</a></h2>
+
+<div>
+<div>
+<pre class="source">CaseExpression ::= SimpleCaseExpression | SearchedCaseExpression
+SimpleCaseExpression ::= &lt;CASE&gt; Expression ( &lt;WHEN&gt; Expression &lt;THEN&gt; Expression )+ ( &lt;ELSE&gt; Expression )? &lt;END&gt;
+SearchedCaseExpression ::= &lt;CASE&gt; ( &lt;WHEN&gt; Expression &lt;THEN&gt; Expression )+ ( &lt;ELSE&gt; Expression )? &lt;END&gt;
+</pre></div></div>
+
+<p>In a simple <tt>CASE</tt> expression, the query evaluator searches for the first <tt>WHEN</tt> &#x2026; <tt>THEN</tt> pair in which the <tt>WHEN</tt> expression is equal to the expression following <tt>CASE</tt> and returns the expression following <tt>THEN</tt>. If none of the <tt>WHEN</tt> &#x2026; <tt>THEN</tt> pairs meet this condition, and an <tt>ELSE</tt> branch exists, it returns the <tt>ELSE</tt> expression. Otherwise, <tt>NULL</tt> is returned.</p>
+<p>In a searched CASE expression, the query evaluator searches from left to right until it finds a <tt>WHEN</tt> expression that is evaluated to <tt>TRUE</tt>, and then returns its corresponding <tt>THEN</tt> expression. If no condition is found to be <tt>TRUE</tt>, and an <tt>ELSE</tt> branch exists, it returns the <tt>ELSE</tt> expression. Otherwise, it returns <tt>NULL</tt>.</p>
+<p>The following example illustrates the form of a case expression.</p>
+<div class="section">
+<div class="section">
+<div class="section">
+<h5><a name="Example"></a>Example</h5>
+
+<div>
+<div>
+<pre class="source">CASE (2 &lt; 3) WHEN true THEN &quot;yes&quot; ELSE &quot;no&quot; END
+</pre></div></div>
+</div></div></div>
+<div class="section">
+<h3><a name="Constructors" id="Constructors">Constructors</a></h3>
+
+<div>
+<div>
+<pre class="source">Constructor              ::= ArrayConstructor | MultisetConstructor | ObjectConstructor
+ArrayConstructor         ::= &quot;[&quot; ( Expression ( &quot;,&quot; Expression )* )? &quot;]&quot;
+MultisetConstructor      ::= &quot;{{&quot; ( Expression ( &quot;,&quot; Expression )* )? &quot;}}&quot;
+ObjectConstructor        ::= &quot;{&quot; ( FieldBinding ( &quot;,&quot; FieldBinding )* )? &quot;}&quot;
+FieldBinding             ::= Expression ( &quot;:&quot; Expression )?
+</pre></div></div>
+
+<p>A major feature of the query language is its ability to construct new data model instances. This is accomplished using its constructors for each of the model&#x2019;s complex object structures, namely arrays, multisets, and objects. Arrays are like JSON arrays, while multisets have bag semantics. Objects are built from fields that are field-name/field-value pairs, again like JSON.</p>
+<p>The following examples illustrate how to construct a new array with 4 items and a new object with 2 fields respectively. Array elements can be homogeneous (as in the first example), which is the common case, or they may be heterogeneous (as in the second example). The data values and field name values used to construct arrays, multisets, and objects in constructors are all simply query expressions. Thus, the collection elements, field names, and field values used in constructors can be simple literals or they can come from query variable references or even arbitrarily complex query expressions (subqueries). Type errors will be raised if the field names in an object are not strings, and duplicate field errors will be raised if they are not distinct.</p>
+<div class="section">
+<div class="section">
+<h5><a name="Examples"></a>Examples</h5>
+
+<div>
+<div>
+<pre class="source">[ 'a', 'b', 'c', 'c' ]
+
+[ 42, &quot;forty-two!&quot;, { &quot;rank&quot; : &quot;Captain&quot;, &quot;name&quot;: &quot;America&quot; }, 3.14159 ]
+
+{
+  'project name': 'Hyracks',
+  'project members': [ 'vinayakb', 'dtabass', 'chenli', 'tsotras', 'tillw' ]
+}
+</pre></div></div>
+
+<p>If only one expression is specified instead of the field-name/field-value pair in an object constructor then this expression is supposed to provide the field value. The field name is then automatically generated based on the kind of the value expression:</p>
+<ul>
+
+<li>If it is a variable reference expression then generated field name is the name of that variable.</li>
+<li>If it is a field access expression then generated field name is the last identifier in that expression.</li>
+<li>For all other cases, a compilation error will be raised.</li>
+</ul></div>
+<div class="section">
+<h5><a name="Example"></a>Example</h5>
+
+<div>
+<div>
+<pre class="source">SELECT VALUE { user.alias, user.userSince }
+FROM GleambookUsers user
+WHERE user.id = 1;
+</pre></div></div>
+
+<p>This query outputs:</p>
+
+<div>
+<div>
+<pre class="source">[ {
+    &quot;alias&quot;: &quot;Margarita&quot;,
+    &quot;userSince&quot;: &quot;2012-08-20T10:10:00&quot;
+} ]
+</pre></div></div>
+<!--
+ ! 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.
+ !-->
+
+<h1><a name="Queries" id="Queries">3. Queries</a></h1>
+<p>A query can be any legal expression or <tt>SELECT</tt> statement. A query always ends with a semicolon.</p>
+
+<div>
+<div>
+<pre class="source">Query ::= (Expression | SelectStatement) &quot;;&quot;
+</pre></div></div>
+<!--
+ ! 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.
+ !-->
+</div></div></div></div>
+<div class="section">
+<h2><a name="Declarations" id="Declarations">Declarations</a></h2>
+
+<div>
+<div>
+<pre class="source">DatabaseDeclaration ::= &quot;USE&quot; Identifier
+</pre></div></div>
+
+<p>At the uppermost level, the world of data is organized into data namespaces called <b>dataverses</b>. To set the default dataverse for statements, the USE statement is provided.</p>
+<p>As an example, the following statement sets the default dataverse to be &#x201c;TinySocial&#x201d;.</p>
+<div class="section">
+<div class="section">
+<div class="section">
+<h5><a name="Example"></a>Example</h5>
+
+<div>
+<div>
+<pre class="source">USE TinySocial;
+</pre></div></div>
+<!--
+ ! 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.
+ !-->
+
+<p>When writing a complex query, it can sometimes be helpful to define one or more auxilliary functions that each address a sub-piece of the overall query. The declare function statement supports the creation of such helper functions. In general, the function body (expression) can be any legal query expression.</p>
+
+<div>
+<div>
+<pre class="source">FunctionDeclaration  ::= &quot;DECLARE&quot; &quot;FUNCTION&quot; Identifier ParameterList &quot;{&quot; Expression &quot;}&quot;
+ParameterList        ::= &quot;(&quot; ( &lt;VARIABLE&gt; ( &quot;,&quot; &lt;VARIABLE&gt; )* )? &quot;)&quot;
+</pre></div></div>
+
+<p>The following is a simple example of a temporary function definition and its use.</p></div>
+<div class="section">
+<h5><a name="Example"></a>Example</h5>
+
+<div>
+<div>
+<pre class="source">DECLARE FUNCTION friendInfo(userId) {
+    (SELECT u.id, u.name, len(u.friendIds) AS friendCount
+     FROM GleambookUsers u
+     WHERE u.id = userId)[0]
+ };
+
+SELECT VALUE friendInfo(2);
+</pre></div></div>
+
+<p>For our sample data set, this returns:</p>
+
+<div>
+<div>
+<pre class="source">[
+  { &quot;id&quot;: 2, &quot;name&quot;: &quot;IsbelDull&quot;, &quot;friendCount&quot;: 2 }
+]
+</pre></div></div>
+<!--
+ ! 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.
+ !-->
+</div></div></div></div>
+<div class="section">
+<h2><a name="SELECT_Statements"></a><a name="SELECT_statements" id="SELECT_statements">SELECT Statements</a></h2>
+<p>The following shows the (rich) grammar for the <tt>SELECT</tt> statement in the query language.</p>
+
+<div>
+<div>
+<pre class="source">SelectStatement    ::= ( WithClause )?
+                       SelectSetOperation (OrderbyClause )? ( LimitClause )?
+SelectSetOperation ::= SelectBlock (&lt;UNION&gt; &lt;ALL&gt; ( SelectBlock | Subquery ) )*
+Subquery           ::= &quot;(&quot; SelectStatement &quot;)&quot;
+
+SelectBlock        ::= SelectClause
+                       ( FromClause ( LetClause )?)?
+                       ( WhereClause )?
+                       ( GroupbyClause ( LetClause )? ( HavingClause )? )?
+                       |
+                       FromClause ( LetClause )?
+                       ( WhereClause )?
+                       ( GroupbyClause ( LetClause )? ( HavingClause )? )?
+                       SelectClause
+
+SelectClause       ::= &lt;SELECT&gt; ( &lt;ALL&gt; | &lt;DISTINCT&gt; )? ( SelectRegular | SelectValue )
+SelectRegular      ::= Projection ( &quot;,&quot; Projection )*
+SelectValue        ::= ( &lt;VALUE&gt; | &lt;ELEMENT&gt; | &lt;RAW&gt; ) Expression
+Projection         ::= ( Expression ( &lt;AS&gt; )? Identifier | &quot;*&quot; | Identifier &quot;.&quot; &quot;*&quot; )
+
+FromClause         ::= &lt;FROM&gt; FromTerm ( &quot;,&quot; FromTerm )*
+FromTerm           ::= Expression (( &lt;AS&gt; )? Variable)?
+                       ( ( JoinType )? ( JoinClause | UnnestClause ) )*
+
+JoinClause         ::= &lt;JOIN&gt; Expression (( &lt;AS&gt; )? Variable)? &lt;ON&gt; Expression
+UnnestClause       ::= ( &lt;UNNEST&gt; ) Expression
+                       ( &lt;AS&gt; )? Variable ( &lt;AT&gt; Variable )?
+JoinType           ::= ( &lt;INNER&gt; | &lt;LEFT&gt; ( &lt;OUTER&gt; )? )
+
+WithClause         ::= &lt;WITH&gt; WithElement ( &quot;,&quot; WithElement )*
+LetClause          ::= (&lt;LET&gt; | &lt;LETTING&gt;) LetElement ( &quot;,&quot; LetElement )*
+LetElement         ::= Variable &quot;=&quot; Expression
+WithElement        ::= Variable &lt;AS&gt; Expression
+
+WhereClause        ::= &lt;WHERE&gt; Expression
+
+GroupbyClause      ::= &lt;GROUP&gt; &lt;BY&gt; Expression ( ( (&lt;AS&gt;)? Variable )?
+                       ( &quot;,&quot; Expression ( (&lt;AS&gt;)? Variable )? )* )
+                       ( &lt;GROUP&gt; &lt;AS&gt; Variable
+                         (&quot;(&quot; VariableReference &lt;AS&gt; Identifier
+                         (&quot;,&quot; VariableReference &lt;AS&gt; Identifier )* &quot;)&quot;)?
+                       )?
+HavingClause       ::= &lt;HAVING&gt; Expression
+
+OrderbyClause      ::= &lt;ORDER&gt; &lt;BY&gt; Expression ( &lt;ASC&gt; | &lt;DESC&gt; )?
+                       ( &quot;,&quot; Expression ( &lt;ASC&gt; | &lt;DESC&gt; )? )*
+LimitClause        ::= &lt;LIMIT&gt; Expression ( &lt;OFFSET&gt; Expression )?
+</pre></div></div>
+
+<p>In this section, we will make use of two stored collections of objects (datasets), <tt>GleambookUsers</tt> and <tt>GleambookMessages</tt>, in a series of running examples to explain <tt>SELECT</tt> queries. The contents of the example collections are as follows:</p>
+<p><tt>GleambookUsers</tt> collection (or, dataset):</p>
+
+<div>
+<div>
+<pre class="source">[ {
+  &quot;id&quot;:1,
+  &quot;alias&quot;:&quot;Margarita&quot;,
+  &quot;name&quot;:&quot;MargaritaStoddard&quot;,
+  &quot;nickname&quot;:&quot;Mags&quot;,
+  &quot;userSince&quot;:&quot;2012-08-20T10:10:00&quot;,
+  &quot;friendIds&quot;:[2,3,6,10],
+  &quot;employment&quot;:[{
+                  &quot;organizationName&quot;:&quot;Codetechno&quot;,
+                  &quot;start-date&quot;:&quot;2006-08-06&quot;
+                },
+                {
+                  &quot;organizationName&quot;:&quot;geomedia&quot;,
+                  &quot;start-date&quot;:&quot;2010-06-17&quot;,
+                  &quot;end-date&quot;:&quot;2010-01-26&quot;
+                }],
+  &quot;gender&quot;:&quot;F&quot;
+},
+{
+  &quot;id&quot;:2,
+  &quot;alias&quot;:&quot;Isbel&quot;,
+  &quot;name&quot;:&quot;IsbelDull&quot;,
+  &quot;nickname&quot;:&quot;Izzy&quot;,
+  &quot;userSince&quot;:&quot;2011-01-22T10:10:00&quot;,
+  &quot;friendIds&quot;:[1,4],
+  &quot;employment&quot;:[{
+                  &quot;organizationName&quot;:&quot;Hexviafind&quot;,
+                  &quot;startDate&quot;:&quot;2010-04-27&quot;
+               }]
+},
+{
+  &quot;id&quot;:3,
+  &quot;alias&quot;:&quot;Emory&quot;,
+  &quot;name&quot;:&quot;EmoryUnk&quot;,
+  &quot;userSince&quot;:&quot;2012-07-10T10:10:00&quot;,
+  &quot;friendIds&quot;:[1,5,8,9],
+  &quot;employment&quot;:[{
+                  &quot;organizationName&quot;:&quot;geomedia&quot;,
+                  &quot;startDate&quot;:&quot;2010-06-17&quot;,
+                  &quot;endDate&quot;:&quot;2010-01-26&quot;
+               }]
+} ]
+</pre></div></div>
+
+<p><tt>GleambookMessages</tt> collection (or, dataset):</p>
+
+<div>
+<div>
+<pre class="source">[ {
+  &quot;messageId&quot;:2,
+  &quot;authorId&quot;:1,
+  &quot;inResponseTo&quot;:4,
+  &quot;senderLocation&quot;:[41.66,80.87],
+  &quot;message&quot;:&quot; dislike x-phone its touch-screen is horrible&quot;
+},
+{
+  &quot;messageId&quot;:3,
+  &quot;authorId&quot;:2,
+  &quot;inResponseTo&quot;:4,
+  &quot;senderLocation&quot;:[48.09,81.01],
+  &quot;message&quot;:&quot; like product-y the plan is amazing&quot;
+},
+{
+  &quot;messageId&quot;:4,
+  &quot;authorId&quot;:1,
+  &quot;inResponseTo&quot;:2,
+  &quot;senderLocation&quot;:[37.73,97.04],
+  &quot;message&quot;:&quot; can't stand acast the network is horrible:(&quot;
+},
+{
+  &quot;messageId&quot;:6,
+  &quot;authorId&quot;:2,
+  &quot;inResponseTo&quot;:1,
+  &quot;senderLocation&quot;:[31.5,75.56],
+  &quot;message&quot;:&quot; like product-z its platform is mind-blowing&quot;
+}
+{
+  &quot;messageId&quot;:8,
+  &quot;authorId&quot;:1,
+  &quot;inResponseTo&quot;:11,
+  &quot;senderLocation&quot;:[40.33,80.87],
+  &quot;message&quot;:&quot; like ccast the 3G is awesome:)&quot;
+},
+{
+  &quot;messageId&quot;:10,
+  &quot;authorId&quot;:1,
+  &quot;inResponseTo&quot;:12,
+  &quot;senderLocation&quot;:[42.5,70.01],
+  &quot;message&quot;:&quot; can't stand product-w the touch-screen is terrible&quot;
+},
+{
+  &quot;messageId&quot;:11,
+  &quot;authorId&quot;:1,
+  &quot;inResponseTo&quot;:1,
+  &quot;senderLocation&quot;:[38.97,77.49],
+  &quot;message&quot;:&quot; can't stand acast its plan is terrible&quot;
+} ]
+</pre></div></div>
+</div>
+<div class="section">
+<h2><a name="SELECT_Clause"></a><a name="Select_clauses" id="Select_clauses">SELECT Clause</a></h2>
+<p>The <tt>SELECT</tt> clause always returns a collection value as its result (even if the result is empty or a singleton).</p>
+<div class="section">
+<h3><a name="Select_Element.2FValue.2FRaw"></a><a name="Select_element" id="Select_element">Select Element/Value/Raw</a></h3>
+<p>The <tt>SELECT VALUE</tt> clause returns an array or multiset that contains the results of evaluating the <tt>VALUE</tt> expression, with one evaluation being performed per &#x201c;binding tuple&#x201d; (i.e., per <tt>FROM</tt> clause item) satisfying the statement&#x2019;s selection criteria. For historical reasons the query language also allows the keywords <tt>ELEMENT</tt> or <tt>RAW</tt> to be used in place of <tt>VALUE</tt> (not recommended).</p>
+<p>If there is no FROM clause, the expression after <tt>VALUE</tt> is evaluated once with no binding tuples (except those inherited from an outer environment).</p>
+<div class="section">
+<div class="section">
+<h5><a name="Example"></a>Example</h5>
+
+<div>
+<div>
+<pre class="source">SELECT VALUE 1;
+</pre></div></div>
+
+<p>This query returns:</p>
+
+<div>
+<div>
+<pre class="source">[
+  1
+]
+</pre></div></div>
+
+<p>The following example shows a query that selects one user from the GleambookUsers collection.</p></div>
+<div class="section">
+<h5><a name="Example"></a>Example</h5>
+
+<div>
+<div>
+<pre class="source">SELECT VALUE user
+FROM GleambookUsers user
+WHERE user.id = 1;
+</pre></div></div>
+
+<p>This query returns:</p>
+
+<div>
+<div>
+<pre class="source">[{
+    &quot;userSince&quot;: &quot;2012-08-20T10:10:00.000Z&quot;,
+    &quot;friendIds&quot;: [
+        2,
+        3,
+        6,
+        10
+    ],
+    &quot;gender&quot;: &quot;F&quot;,
+    &quot;name&quot;: &quot;MargaritaStoddard&quot;,
+    &quot;nickname&quot;: &quot;Mags&quot;,
+    &quot;alias&quot;: &quot;Margarita&quot;,
+    &quot;id&quot;: 1,
+    &quot;employment&quot;: [
+        {
+            &quot;organizationName&quot;: &quot;Codetechno&quot;,
+            &quot;start-date&quot;: &quot;2006-08-06&quot;
+        },
+        {
+            &quot;end-date&quot;: &quot;2010-01-26&quot;,
+            &quot;organizationName&quot;: &quot;geomedia&quot;,
+            &quot;start-date&quot;: &quot;2010-06-17&quot;
+        }
+    ]
+} ]
+</pre></div></div>
+</div></div></div>
+<div class="section">
+<h3><a name="SQL-style_SELECT"></a><a name="SQL_select" id="SQL_select">SQL-style SELECT</a></h3>
+<p>The traditional SQL-style <tt>SELECT</tt> syntax is also supported in the query language. This syntax can also be reformulated in a <tt>SELECT VALUE</tt> based manner. (E.g., <tt>SELECT expA AS fldA, expB AS fldB</tt> is syntactic sugar for <tt>SELECT VALUE { 'fldA': expA, 'fldB': expB }</tt>.) Unlike in SQL, the result of a query does not preserve the order of expressions in the <tt>SELECT</tt> clause.</p>
+<div class="section">
+<div class="section">
+<h5><a name="Example"></a>Example</h5>
+
+<div>
+<div>
+<pre class="source">SELECT user.alias user_alias, user.name user_name
+FROM GleambookUsers user
+WHERE user.id = 1;
+</pre></div></div>
+
+<p>Returns:</p>
+
+<div>
+<div>
+<pre class="source">[ {
+    &quot;user_name&quot;: &quot;MargaritaStoddard&quot;,
+    &quot;user_alias&quot;: &quot;Margarita&quot;
+} ]
+</pre></div></div>
+</div></div></div>
+<div class="section">
+<h3><a name="SELECT_.2A"></a><a name="Select_star" id="Select_star">SELECT *</a></h3>
+<p><tt>SELECT *</tt> returns an object with a nested field for each input tuple. Each field has as its field name the name of a binding variable generated by either the <tt>FROM</tt> clause or <tt>GROUP BY</tt> clause in the current enclosing <tt>SELECT</tt> statement, and its field value is the value of that binding variable.</p>
+<p>Note that the result of <tt>SELECT *</tt> is different from the result of query that selects all the fields of an object.</p>
+<div class="section">
+<div class="section">
+<h5><a name="Example"></a>Example</h5>
+
+<div>
+<div>
+<pre class="source">SELECT *
+FROM GleambookUsers user;
+</pre></div></div>
+
+<p>Since <tt>user</tt> is the only binding variable generated in the <tt>FROM</tt> clause, this query returns:</p>
+
+<div>
+<div>
+<pre class="source">[ {
+    &quot;user&quot;: {
+        &quot;userSince&quot;: &quot;2012-08-20T10:10:00.000Z&quot;,
+        &quot;friendIds&quot;: [
+            2,
+            3,
+            6,
+            10
+        ],
+        &quot;gender&quot;: &quot;F&quot;,
+        &quot;name&quot;: &quot;MargaritaStoddard&quot;,
+        &quot;nickname&quot;: &quot;Mags&quot;,
+        &quot;alias&quot;: &quot;Margarita&quot;,
+        &quot;id&quot;: 1,
+        &quot;employment&quot;: [
+            {
+                &quot;organizationName&quot;: &quot;Codetechno&quot;,
+                &quot;start-date&quot;: &quot;2006-08-06&quot;
+            },
+            {
+                &quot;end-date&quot;: &quot;2010-01-26&quot;,
+                &quot;organizationName&quot;: &quot;geomedia&quot;,
+                &quot;start-date&quot;: &quot;2010-06-17&quot;
+            }
+        ]
+    }
+}, {
+    &quot;user&quot;: {
+        &quot;userSince&quot;: &quot;2011-01-22T10:10:00.000Z&quot;,
+        &quot;friendIds&quot;: [
+            1,
+            4
+        ],
+        &quot;name&quot;: &quot;IsbelDull&quot;,
+        &quot;nickname&quot;: &quot;Izzy&quot;,
+        &quot;alias&quot;: &quot;Isbel&quot;,
+        &quot;id&quot;: 2,
+        &quot;employment&quot;: [
+            {
+                &quot;organizationName&quot;: &quot;Hexviafind&quot;,
+                &quot;startDate&quot;: &quot;2010-04-27&quot;
+            }
+        ]
+    }
+}, {
+    &quot;user&quot;: {
+        &quot;userSince&quot;: &quot;2012-07-10T10:10:00.000Z&quot;,
+        &quot;friendIds&quot;: [
+            1,
+            5,
+            8,
+            9
+        ],
+        &quot;name&quot;: &quot;EmoryUnk&quot;,
+        &quot;alias&quot;: &quot;Emory&quot;,
+        &quot;id&quot;: 3,
+        &quot;employment&quot;: [
+            {
+                &quot;organizationName&quot;: &quot;geomedia&quot;,
+                &quot;endDate&quot;: &quot;2010-01-26&quot;,
+                &quot;startDate&quot;: &quot;2010-06-17&quot;
+            }
+        ]
+    }
+} ]
+</pre></div></div>
+</div>
+<div class="section">
+<h5><a name="Example"></a>Example</h5>
+
+<div>
+<div>
+<pre class="source">SELECT *
+FROM GleambookUsers u, GleambookMessages m
+WHERE m.authorId = u.id and u.id = 2;
+</pre></div></div>
+
+<p>This query does an inner join that we will discuss in <a href="#Multiple_from_terms">multiple from terms</a>. Since both <tt>u</tt> and <tt>m</tt> are binding variables generated in the <tt>FROM</tt> clause, this query returns:</p>
+
+<div>
+<div>
+<pre class="source">[ {
+    &quot;u&quot;: {
+        &quot;userSince&quot;: &quot;2011-01-22T10:10:00&quot;,
+        &quot;friendIds&quot;: [
+            1,
+            4
+        ],
+        &quot;name&quot;: &quot;IsbelDull&quot;,
+        &quot;nickname&quot;: &quot;Izzy&quot;,
+        &quot;alias&quot;: &quot;Isbel&quot;,
+        &quot;id&quot;: 2,
+        &quot;employment&quot;: [
+            {
+                &quot;organizationName&quot;: &quot;Hexviafind&quot;,
+                &quot;startDate&quot;: &quot;2010-04-27&quot;
+            }
+        ]
+    },
+    &quot;m&quot;: {
+        &quot;senderLocation&quot;: [
+            31.5,
+            75.56
+        ],
+        &quot;inResponseTo&quot;: 1,
+        &quot;messageId&quot;: 6,
+        &quot;authorId&quot;: 2,
+        &quot;message&quot;: &quot; like product-z its platform is mind-blowing&quot;
+    }
+}, {
+    &quot;u&quot;: {
+        &quot;userSince&quot;: &quot;2011-01-22T10:10:00&quot;,
+        &quot;friendIds&quot;: [
+            1,
+            4
+        ],
+        &quot;name&quot;: &quot;IsbelDull&quot;,
+        &quot;nickname&quot;: &quot;Izzy&quot;,
+        &quot;alias&quot;: &quot;Isbel&quot;,
+        &quot;id&quot;: 2,
+        &quot;employment&quot;: [
+            {
+                &quot;organizationName&quot;: &quot;Hexviafind&quot;,
+                &quot;startDate&quot;: &quot;2010-04-27&quot;
+            }
+        ]
+    },
+    &quot;m&quot;: {
+        &quot;senderLocation&quot;: [
+            48.09,
+            81.01
+        ],
+        &quot;inResponseTo&quot;: 4,
+        &quot;messageId&quot;: 3,
+        &quot;authorId&quot;: 2,
+        &quot;message&quot;: &quot; like product-y the plan is amazing&quot;
+    }
+} ]
+</pre></div></div>
+</div></div></div>
+<div class="section">
+<h3><a name="SELECT_variable..2A"></a><a name="Select_variable_star" id="Select_variable_star">SELECT <i>variable</i>.*</a></h3>
+<p>Whereas <tt>SELECT *</tt> returns all the fields bound to all the variables which are currently defined, the notation <tt>SELECT c.*</tt> returns all the fields of the object bound to variable <tt>c</tt>. The variable <tt>c</tt> must be bound to an object for this to work.</p>
+<div class="section">
+<div class="section">
+<h5><a name="Example"></a>Example</h5>
+
+<div>
+<div>
+<pre class="source">SELECT user.*
+FROM GleambookUsers user;
+</pre></div></div>
+
+<p>Compare this query with the first example given under <a href="#Select_star">SELECT *</a>. This query returns all users from the <tt>GleambookUsers</tt> dataset, but the <tt>user</tt> variable name is omitted from the results:</p>
+
+<div>
+<div>
+<pre class="source">[
+  {
+    &quot;id&quot;: 1,
+    &quot;alias&quot;: &quot;Margarita&quot;,
+    &quot;name&quot;: &quot;MargaritaStoddard&quot;,
+    &quot;nickname&quot;: &quot;Mags&quot;,
+    &quot;userSince&quot;: &quot;2012-08-20T10:10:00&quot;,
+    &quot;friendIds&quot;: [
+      2,
+      3,
+      6,
+      10
+    ],
+    &quot;employment&quot;: [
+      {
+        &quot;organizationName&quot;: &quot;Codetechno&quot;,
+        &quot;start-date&quot;: &quot;2006-08-06&quot;
+      },
+      {
+        &quot;organizationName&quot;: &quot;geomedia&quot;,
+        &quot;start-date&quot;: &quot;2010-06-17&quot;,
+        &quot;end-date&quot;: &quot;2010-01-26&quot;
+      }
+    ],
+    &quot;gender&quot;: &quot;F&quot;
+  },
+  {
+    &quot;id&quot;: 2,
+    &quot;alias&quot;: &quot;Isbel&quot;,
+    &quot;name&quot;: &quot;IsbelDull&quot;,
+    &quot;nickname&quot;: &quot;Izzy&quot;,
+    &quot;userSince&quot;: &quot;2011-01-22T10:10:00&quot;,
+    &quot;friendIds&quot;: [
+      1,
+      4
+    ],
+    &quot;employment&quot;: [
+      {
+        &quot;organizationName&quot;: &quot;Hexviafind&quot;,
+        &quot;startDate&quot;: &quot;2010-04-27&quot;
+      }
+    ]
+  },
+  {
+    &quot;id&quot;: 3,
+    &quot;alias&quot;: &quot;Emory&quot;,
+    &quot;name&quot;: &quot;EmoryUnk&quot;,
+    &quot;userSince&quot;: &quot;2012-07-10T10:10:00&quot;,
+    &quot;friendIds&quot;: [
+      1,
+      5,
+      8,
+      9
+    ],
+    &quot;employment&quot;: [
+      {
+        &quot;organizationName&quot;: &quot;geomedia&quot;,
+        &quot;startDate&quot;: &quot;2010-06-17&quot;,
+        &quot;endDate&quot;: &quot;2010-01-26&quot;
+      }
+    ]
+  }
+]
+</pre></div></div>
+</div></div></div>
+<div class="section">
+<h3><a name="SELECT_DISTINCT"></a><a name="Select_distinct" id="Select_distinct">SELECT DISTINCT</a></h3>
+<p>The <tt>DISTINCT</tt> keyword is used to eliminate duplicate items in results. The following example shows how it works.</p>
+<div class="section">
+<div class="section">
+<h5><a name="Example"></a>Example</h5>
+
+<div>
+<div>
+<pre class="source">SELECT DISTINCT * FROM [1, 2, 2, 3] AS foo;
+</pre></div></div>
+
+<p>This query returns:</p>
+
+<div>
+<div>
+<pre class="source">[ {
+    &quot;foo&quot;: 1
+}, {
+    &quot;foo&quot;: 2
+}, {
+    &quot;foo&quot;: 3
+} ]
+</pre></div></div>
+</div>
+<div class="section">
+<h5><a name="Example"></a>Example</h5>
+
+<div>
+<div>
+<pre class="source">SELECT DISTINCT VALUE foo FROM [1, 2, 2, 3] AS foo;
+</pre></div></div>
+
+<p>This version of the query returns:</p>
+
+<div>
+<div>
+<pre class="source">[ 1
+, 2
+, 3
+ ]
+</pre></div></div>
+</div></div></div>
+<div class="section">
+<h3><a name="Unnamed_Projections"></a><a name="Unnamed_projections" id="Unnamed_projections">Unnamed Projections</a></h3>
+<p>Similar to standard SQL, the query language  supports unnamed projections (a.k.a, unnamed <tt>SELECT</tt> clause items), for which names are generated. Name generation has three cases:</p>
+<ul>
+
+<li>If a projection expression is a variable reference expression, its generated name is the name of the variable.</li>
+<li>If a projection expression is a field access expression, its generated name is the last identifier in the expression.</li>
+<li>For all other cases, the query processor will generate a unique name.</li>
+</ul>
+<div class="section">
+<div class="section">
+<h5><a name="Example"></a>Example</h5>
+
+<div>
+<div>
+<pre class="source">SELECT substr(user.name, 10), user.alias
+FROM GleambookUsers user
+WHERE user.id = 1;
+</pre></div></div>
+
+<p>This query outputs:</p>
+
+<div>
+<div>
+<pre class="source">[ {
+    &quot;alias&quot;: &quot;Margarita&quot;,
+    &quot;$1&quot;: &quot;Stoddard&quot;
+} ]
+</pre></div></div>
+
+<p>In the result, <tt>$1</tt> is the generated name for <tt>substr(user.name, 1)</tt>, while <tt>alias</tt> is the generated name for <tt>user.alias</tt>.</p></div></div></div>
+<div class="section">
+<h3><a name="Abbreviated_Field_Access_Expressions"></a><a name="Abbreviated_field_access_expressions" id="Abbreviated_field_access_expressions">Abbreviated Field Access Expressions</a></h3>
+<p>As in standard SQL, field access expressions can be abbreviated (not recommended!) when there is no ambiguity. In the next example, the variable <tt>user</tt> is the only possible variable reference for fields <tt>id</tt>, <tt>name</tt> and <tt>alias</tt> and thus could be omitted in the query. More information on abbbreviated field access can be found in the appendix section on Variable Resolution.</p>
+<div class="section">
+<div class="section">
+<h5><a name="Example"></a>Example</h5>
+
+<div>
+<div>
+<pre class="source">SELECT substr(name, 10) AS lname, alias
+FROM GleambookUsers user
+WHERE id = 1;
+</pre></div></div>
+
+<p>Outputs:</p>
+
+<div>
+<div>
+<pre class="source">[ {
+    &quot;lname&quot;: &quot;Stoddard&quot;,
+    &quot;alias&quot;: &quot;Margarita&quot;
+} ]
+</pre></div></div>
+</div></div></div></div>
+<div class="section">
+<h2><a name="UNNEST_Clause"></a><a name="Unnest_clauses" id="Unnest_clauses">UNNEST Clause</a></h2>
+<p>For each of its input tuples, the <tt>UNNEST</tt> clause flattens a collection-valued expression into individual items, producing multiple tuples, each of which is one of the expression&#x2019;s original input tuples augmented with a flattened item from its collection.</p>
+<div class="section">
+<h3><a name="Inner_UNNEST"></a><a name="Inner_unnests" id="Inner_unnests">Inner UNNEST</a></h3>
+<p>The following example is a query that retrieves the names of the organizations that a selected user has worked for. It uses the <tt>UNNEST</tt> clause to unnest the nested collection <tt>employment</tt> in the user&#x2019;s object.</p>
+<div class="section">
+<div class="section">
+<h5><a name="Example"></a>Example</h5>
+
+<div>
+<div>
+<pre class="source">SELECT u.id AS userId, e.organizationName AS orgName
+FROM GleambookUsers u
+UNNEST u.employment e
+WHERE u.id = 1;
+</pre></div></div>
+
+<p>This query returns:</p>
+
+<div>
+<div>
+<pre class="source">[ {
+    &quot;orgName&quot;: &quot;Codetechno&quot;,
+    &quot;userId&quot;: 1
+}, {
+    &quot;orgName&quot;: &quot;geomedia&quot;,
+    &quot;userId&quot;: 1
+} ]
+</pre></div></div>
+
+<p>Note that <tt>UNNEST</tt> has SQL&#x2019;s inner join semantics &#x2014; that is, if a user has no employment history, no tuple corresponding to that user will be emitted in the result.</p></div></div></div>
+<div class="section">
+<h3><a name="Left_Outer_UNNEST"></a><a name="Left_outer_unnests" id="Left_outer_unnests">Left Outer UNNEST</a></h3>
+<p>As an alternative, the <tt>LEFT OUTER UNNEST</tt> clause offers SQL&#x2019;s left outer join semantics. For example, no collection-valued field named <tt>hobbies</tt> exists in the object for the user whose id is 1, but the following query&#x2019;s result still includes user 1.</p>
+<div class="section">
+<div class="section">
+<h5><a name="Example"></a>Example</h5>
+
+<div>
+<div>
+<pre class="source">SELECT u.id AS userId, h.hobbyName AS hobby
+FROM GleambookUsers u
+LEFT OUTER UNNEST u.hobbies h
+WHERE u.id = 1;
+</pre></div></div>
+
+<p>Returns:</p>
+
+<div>
+<div>
+<pre class="source">[ {
+    &quot;userId&quot;: 1
+} ]
+</pre></div></div>
+
+<p>Note that if <tt>u.hobbies</tt> is an empty collection or leads to a <tt>MISSING</tt> (as above) or <tt>NULL</tt> value for a given input tuple, there is no corresponding binding value for variable <tt>h</tt> for an input tuple. A <tt>MISSING</tt> value will be generated for <tt>h</tt> so that the input tuple can still be propagated.</p></div></div></div>
+<div class="section">
+<h3><a name="Expressing_Joins_Using_UNNEST"></a><a name="Expressing_joins_using_unnests" id="Expressing_joins_using_unnests">Expressing Joins Using UNNEST</a></h3>
+<p>The <tt>UNNEST</tt> clause is similar to SQL&#x2019;s <tt>JOIN</tt> clause except that it allows its right argument to be correlated to its left argument, as in the examples above &#x2014; i.e., think &#x201c;correlated cross-product&#x201d;. The next example shows this via a query that joins two data sets, GleambookUsers and GleambookMessages, returning user/message pairs. The results contain one object per pair, with result objects containing the user&#x2019;s name and an entire message. The query can be thought of as saying &#x201c;for each Gleambook user, unnest the <tt>GleambookMessages</tt> collection and filter the output with the condition <tt>message.authorId = user.id</tt>&#x201d;.</p>
+<div class="section">
+<div class="section">
+<h5><a name="Example"></a>Example</h5>
+
+<div>
+<div>
+<pre class="source">SELECT u.name AS uname, m.message AS message
+FROM GleambookUsers u
+UNNEST GleambookMessages m
+WHERE m.authorId = u.id;
+</pre></div></div>
+
+<p>This returns:</p>
+
+<div>
+<div>
+<pre class="source">[ {
+    &quot;uname&quot;: &quot;MargaritaStoddard&quot;,
+    &quot;message&quot;: &quot; can't stand acast its plan is terrible&quot;
+}, {
+    &quot;uname&quot;: &quot;MargaritaStoddard&quot;,
+    &quot;message&quot;: &quot; dislike x-phone its touch-screen is horrible&quot;
+}, {
+    &quot;uname&quot;: &quot;MargaritaStoddard&quot;,
+    &quot;message&quot;: &quot; can't stand acast the network is horrible:(&quot;
+}, {
+    &quot;uname&quot;: &quot;MargaritaStoddard&quot;,
+    &quot;message&quot;: &quot; like ccast the 3G is awesome:)&quot;
+}, {
+    &quot;uname&quot;: &quot;MargaritaStoddard&quot;,
+    &quot;message&quot;: &quot; can't stand product-w the touch-screen is terrible&quot;
+}, {
+    &quot;uname&quot;: &quot;IsbelDull&quot;,
+    &quot;message&quot;: &quot; like product-z its platform is mind-blowing&quot;
+}, {
+    &quot;uname&quot;: &quot;IsbelDull&quot;,
+    &quot;message&quot;: &quot; like product-y the plan is amazing&quot;
+} ]
+</pre></div></div>
+
+<p>Similarly, the above query can also be expressed as the <tt>UNNEST</tt>ing of a correlated subquery:</p></div>
+<div class="section">
+<h5><a name="Example"></a>Example</h5>
+
+<div>
+<div>
+<pre class="source">SELECT u.name AS uname, m.message AS message
+FROM GleambookUsers u
+UNNEST (
+    SELECT VALUE msg
+    FROM GleambookMessages msg
+    WHERE msg.authorId = u.id
+) AS m;
+</pre></div></div>
+</div></div></div></div>
+<div class="section">
+<h2><a name="FROM_clauses"></a><a name="From_clauses" id="From_clauses">FROM clauses</a></h2>
+<p>A <tt>FROM</tt> clause is used for enumerating (i.e., conceptually iterating over) the contents of collections, as in SQL.</p>
+<div class="section">
+<h3><a name="Binding_expressions" id="Binding_expressions">Binding expressions</a></h3>
+<p>In addition to stored collections, a <tt>FROM</tt> clause can iterate over any intermediate collection returned by a valid query expression. In the tuple stream generated by a <tt>FROM</tt> clause, the ordering of the input tuples are not guaranteed to be preserved.</p>
+<div class="section">
+<div class="section">
+<h5><a name="Example"></a>Example</h5>
+
+<div>
+<div>
+<pre class="source">SELECT VALUE foo
+FROM [1, 2, 2, 3] AS foo
+WHERE foo &gt; 2;
+</pre></div></div>
+
+<p>Returns:</p>
+
+<div>
+<div>
+<pre class="source">[
+  3
+]
+</pre></div></div>
+</div></div></div>
+<div class="section">
+<h3><a name="Multiple_FROM_Terms"></a><a name="Multiple_from_terms" id="Multiple_from_terms">Multiple FROM Terms</a></h3>
+<p>The query language permits correlations among <tt>FROM</tt> terms. Specifically, a <tt>FROM</tt> binding expression can refer to variables defined to its left in the given <tt>FROM</tt> clause. Thus, the first unnesting example above could also be expressed as follows:</p>
+<div class="section">
+<div class="section">
+<h5><a name="Example"></a>Example</h5>
+
+<div>
+<div>
+<pre class="source">SELECT u.id AS userId, e.organizationName AS orgName
+FROM GleambookUsers u, u.employment e
+WHERE u.id = 1;
+</pre></div></div>
+</div></div></div>
+<div class="section">
+<h3><a name="Expressing_Joins_Using_FROM_Terms"></a><a name="Expressing_joins_using_from_terms" id="Expressing_joins_using_from_terms">Expressing Joins Using FROM Terms</a></h3>
+<p>Similarly, the join intentions of the other <tt>UNNEST</tt>-based join examples above could be expressed as:</p>
+<div class="section">
+<div class="section">
+<h5><a name="Example"></a>Example</h5>
+
+<div>
+<div>
+<pre class="source">SELECT u.name AS uname, m.message AS message
+FROM GleambookUsers u, GleambookMessages m
+WHERE m.authorId = u.id;
+</pre></div></div>
+</div>
+<div class="section">
+<h5><a name="Example"></a>Example</h5>
+
+<div>
+<div>
+<pre class="source">SELECT u.name AS uname, m.message AS message
+FROM GleambookUsers u,
+  (
+    SELECT VALUE msg
+    FROM GleambookMessages msg
+    WHERE msg.authorId = u.id
+  ) AS m;
+</pre></div></div>
+
+<p>Note that the first alternative is one of the SQL-92 approaches to expressing a join.</p></div></div></div>
+<div class="section">
+<h3><a name="Implicit_Binding_Variables"></a><a name="Implicit_binding_variables" id="Implicit_binding_variables">Implicit Binding Variables</a></h3>
+<p>Similar to standard SQL, the query language supports implicit <tt>FROM</tt> binding variables (i.e., aliases), for which a binding variable is generated. Variable generation falls into three cases:</p>
+<ul>
+
+<li>If the binding expression is a variable reference expression, the generated variable&#x2019;s name will be the name of the referenced variable itself.</li>
+<li>If the binding expression is a field access expression (or a fully qualified name for a dataset), the generated variable&#x2019;s name will be the last identifier (or the dataset name) in the expression.</li>
+<li>For all other cases, a compilation error will be raised.</li>
+</ul>
+<p>The next two examples show queries that do not provide binding variables in their <tt>FROM</tt> clauses.</p>
+<div class="section">
+<div class="section">
+<h5><a name="Example"></a>Example</h5>
+
+<div>
+<div>
+<pre class="source">SELECT GleambookUsers.name, GleambookMessages.message
+FROM GleambookUsers, GleambookMessages
+WHERE GleambookMessages.authorId = GleambookUsers.id;
+</pre></div></div>
+
+<p>Returns:</p>
+
+<div>
+<div>
+<pre class="source">[ {
+    &quot;name&quot;: &quot;MargaritaStoddard&quot;,
+    &quot;message&quot;: &quot; like ccast the 3G is awesome:)&quot;
+}, {
+    &quot;name&quot;: &quot;MargaritaStoddard&quot;,
+    &quot;message&quot;: &quot; can't stand product-w the touch-screen is terrible&quot;
+}, {
+    &quot;name&quot;: &quot;MargaritaStoddard&quot;,
+    &quot;message&quot;: &quot; can't stand acast its plan is terrible&quot;
+}, {
+    &quot;name&quot;: &quot;MargaritaStoddard&quot;,
+    &quot;message&quot;: &quot; dislike x-phone its touch-screen is horrible&quot;
+}, {
+    &quot;name&quot;: &quot;MargaritaStoddard&quot;,
+    &quot;message&quot;: &quot; can't stand acast the network is horrible:(&quot;
+}, {
+    &quot;name&quot;: &quot;IsbelDull&quot;,
+    &quot;message&quot;: &quot; like product-y the plan is amazing&quot;
+}, {
+    &quot;name&quot;: &quot;IsbelDull&quot;,
+    &quot;message&quot;: &quot; like product-z its platform is mind-blowing&quot;
+} ]
+</pre></div></div>
+</div>
+<div class="section">
+<h5><a name="Example"></a>Example</h5>
+
+<div>
+<div>
+<pre class="source">SELECT GleambookUsers.name, GleambookMessages.message
+FROM GleambookUsers,
+  (
+    SELECT VALUE GleambookMessages
+    FROM GleambookMessages
+    WHERE GleambookMessages.authorId = GleambookUsers.id
+  );
+</pre></div></div>
+
+<p>Returns:</p>
+
+<div>
+<div>
+<pre class="source">Error: &quot;Syntax error: Need an alias for the enclosed expression:\n(select element GleambookMessages\n    from GleambookMessages as GleambookMessages\n    where (GleambookMessages.authorId = GleambookUsers.id)\n )&quot;,
+    &quot;query_from_user&quot;: &quot;use TinySocial;\n\nSELECT GleambookUsers.name, GleambookMessages.message\n    FROM GleambookUsers,\n      (\n        SELECT VALUE GleambookMessages\n        FROM GleambookMessages\n        WHERE GleambookMessages.authorId = GleambookUsers.id\n      );&quot;
+</pre></div></div>
+
+<p>More information on implicit binding variables can be found in the appendix section on Variable Resolution.</p></div></div></div></div>
+<div class="section">
+<h2><a name="JOIN_Clauses"></a><a name="Join_clauses" id="Join_clauses">JOIN Clauses</a></h2>
+<p>The join clause in the query language supports both inner joins and left outer joins from standard SQL.</p>
+<div class="section">
+<h3><a name="Inner_joins" id="Inner_joins">Inner joins</a></h3>
+<p>Using a <tt>JOIN</tt> clause, the inner join intent from the preceding examples can also be expressed as follows:</p>
+<div class="section">
+<div class="section">
+<h5><a name="Example"></a>Example</h5>
+
+<div>
+<div>
+<pre class="source">SELECT u.name AS uname, m.message AS message
+FROM GleambookUsers u JOIN GleambookMessages m ON m.authorId = u.id;
+</pre></div></div>
+</div></div></div>
+<div class="section">
+<h3><a name="Left_Outer_Joins"></a><a name="Left_outer_joins" id="Left_outer_joins">Left Outer Joins</a></h3>
+<p>The query language supports SQL&#x2019;s notion of left outer join. The following query is an example:</p>
+
+<div>
+<div>
+<pre class="source">SELECT u.name AS uname, m.message AS message
+FROM GleambookUsers u LEFT OUTER JOIN GleambookMessages m ON m.authorId = u.id;
+</pre></div></div>
+
+<p>Returns:</p>
+
+<div>
+<div>
+<pre class="source">[ {
+    &quot;uname&quot;: &quot;MargaritaStoddard&quot;,
+    &quot;message&quot;: &quot; like ccast the 3G is awesome:)&quot;
+}, {
+    &quot;uname&quot;: &quot;MargaritaStoddard&quot;,
+    &quot;message&quot;: &quot; can't stand product-w the touch-screen is terrible&quot;
+}, {
+    &quot;uname&quot;: &quot;MargaritaStoddard&quot;,
+    &quot;message&quot;: &quot; can't stand acast its plan is terrible&quot;
+}, {
+    &quot;uname&quot;: &quot;MargaritaStoddard&quot;,
+    &quot;message&quot;: &quot; dislike x-phone its touch-screen is horrible&quot;
+}, {
+    &quot;uname&quot;: &quot;MargaritaStoddard&quot;,
+    &quot;message&quot;: &quot; can't stand acast the network is horrible:(&quot;
+}, {
+    &quot;uname&quot;: &quot;IsbelDull&quot;,
+    &quot;message&quot;: &quot; like product-y the plan is amazing&quot;
+}, {
+    &quot;uname&quot;: &quot;IsbelDull&quot;,
+    &quot;message&quot;: &quot; like product-z its platform is mind-blowing&quot;
+}, {
+    &quot;uname&quot;: &quot;EmoryUnk&quot;
+} ]
+</pre></div></div>
+
+<p>For non-matching left-side tuples, the query language produces <tt>MISSING</tt> values for the right-side binding variables; that is why the last object in the above result doesn&#x2019;t have a <tt>message</tt> field. Note that this is slightly different from standard SQL, which instead would fill in <tt>NULL</tt> values for the right-side fields. The reason for this difference is that, for non-matches in its join results, the query language views fields from the right-side as being &#x201c;not there&#x201d; (a.k.a. <tt>MISSING</tt>) instead of as being &#x201c;there but unknown&#x201d; (i.e., <tt>NULL</tt>).</p>
+<p>The left-outer join query can also be expressed using <tt>LEFT OUTER UNNEST</tt>:</p>
+
+<div>
+<div>
+<pre class="source">SELECT u.name AS uname, m.message AS message
+FROM GleambookUsers u
+LEFT OUTER UNNEST (
+    SELECT VALUE message
+    FROM GleambookMessages message
+    WHERE message.authorId = u.id
+  ) m;
+</pre></div></div>
+
+<p>In general, SQL-style join queries can also be expressed by <tt>UNNEST</tt> clauses and left outer join queries can be expressed by <tt>LEFT OUTER UNNESTs</tt>.</p></div>
+<div class="section">
+<h3><a name="Variable_scope_in_JOIN_clauses"></a><a name="Join_variable_scope" id="Join_variable_scope">Variable scope in JOIN clauses</a></h3>
+<p>Variables defined by <tt>JOIN</tt> subclauses are not visible to other subclauses in the same <tt>FROM</tt> clause. This also applies to the <tt>FROM</tt> variable that starts the <tt>JOIN</tt> subclause.</p>
+<div class="section">
+<div class="section">
+<h5><a name="Example"></a>Example</h5>
+
+<div>
+<div>
+<pre class="source">SELECT * FROM GleambookUsers u
+JOIN (SELECT VALUE m
+      FROM GleambookMessages m
+      WHERE m.authorId = u.id) m
+ON u.id = m.authorId;
+</pre></div></div>
+
+<p>The variable <tt>u</tt> defined by the <tt>FROM</tt> clause is not visible inside the <tt>JOIN</tt> subclause, so this query returns no results.</p></div></div></div></div>
+<div class="section">
+<h2><a name="GROUP_BY_Clauses"></a><a name="Group_By_clauses" id="Group_By_clauses">GROUP BY Clauses</a></h2>
+<p>The <tt>GROUP BY</tt> clause generalizes standard SQL&#x2019;s grouping and aggregation semantics, but it also retains backward compatibility with the standard (relational) SQL <tt>GROUP BY</tt> and aggregation features.</p>
+<div class="section">
+<h3><a name="Group_variables" id="Group_variables">Group variables</a></h3>
+<p>In a <tt>GROUP BY</tt> clause, in addition to the binding variable(s) defined for the grouping key(s), the query language allows a user to define a <i>group variable</i> by using the clause&#x2019;s <tt>GROUP AS</tt> extension to denote the resulting group. After grouping, then, the query&#x2019;s in-scope variables include the grouping key&#x2019;s binding variables as well as this group variable which will be bound to one collection value for each group. This per-group collection (i.e., multiset) value will be a set of nested objects in which each field of the object is the result of a renamed variable defined in parentheses following the group variable&#x2019;s name. The <tt>GROUP AS</tt> syntax is as follows:</p>
+
+<div>
+<div>
+<pre class="source">&lt;GROUP&gt; &lt;AS&gt; Variable (&quot;(&quot; VariableReference &lt;AS&gt; Identifier (&quot;,&quot; VariableReference &lt;AS&gt; Identifier )* &quot;)&quot;)?
+</pre></div></div>
+
+<div class="section">
+<div class="section">
+<h5><a name="Example"></a>Example</h5>
+
+<div>
+<div>
+<pre class="source">SELECT *
+FROM GleambookMessages message
+GROUP BY message.authorId AS uid GROUP AS msgs(message AS msg);
+</pre></div></div>
+
+<p>This first example query returns:</p>
+
+<div>
+<div>
+<pre class="source">[ {
+    &quot;msgs&quot;: [
+        {
+            &quot;msg&quot;: {
+                &quot;senderLocation&quot;: [
+                    38.97,
+                    77.49
+                ],
+                &quot;inResponseTo&quot;: 1,
+                &quot;messageId&quot;: 11,
+                &quot;authorId&quot;: 1,
+                &quot;message&quot;: &quot; can't stand acast its plan is terrible&quot;
+            }
+        },
+        {
+            &quot;msg&quot;: {
+                &quot;senderLocation&quot;: [
+                    41.66,
+                    80.87
+                ],
+                &quot;inResponseTo&quot;: 4,
+                &quot;messageId&quot;: 2,
+                &quot;authorId&quot;: 1,
+                &quot;message&quot;: &quot; dislike x-phone its touch-screen is horrible&quot;
+            }
+        },
+        {
+            &quot;msg&quot;: {
+                &quot;senderLocation&quot;: [
+                    37.73,
+                    97.04
+                ],
+                &quot;inResponseTo&quot;: 2,
+                &quot;messageId&quot;: 4,
+                &quot;authorId&quot;: 1,
+                &quot;message&quot;: &quot; can't stand acast the network is horrible:(&quot;
+            }
+        },
+        {
+            &quot;msg&quot;: {
+                &quot;senderLocation&quot;: [
+                    40.33,
+                    80.87
+                ],
+                &quot;inResponseTo&quot;: 11,
+                &quot;messageId&quot;: 8,
+                &quot;authorId&quot;: 1,
+                &quot;message&quot;: &quot; like ccast the 3G is awesome:)&quot;
+            }
+        },
+        {
+            &quot;msg&quot;: {
+                &quot;senderLocation&quot;: [
+                    42.5,
+                    70.01
+                ],
+                &quot;inResponseTo&quot;: 12,
+                &quot;messageId&quot;: 10,
+                &quot;authorId&quot;: 1,
+                &quot;message&quot;: &quot; can't stand product-w the touch-screen is terrible&quot;
+            }
+        }
+    ],
+    &quot;uid&quot;: 1
+}, {
+    &quot;msgs&quot;: [
+        {
+            &quot;msg&quot;: {
+                &quot;senderLocation&quot;: [
+                    31.5,
+                    75.56
+                ],
+                &quot;inResponseTo&quot;: 1,
+                &quot;messageId&quot;: 6,
+                &quot;authorId&quot;: 2,
+                &quot;message&quot;: &quot; like product-z its platform is mind-blowing&quot;
+            }
+        },
+        {
+            &quot;msg&quot;: {
+                &quot;senderLocation&quot;: [
+                    48.09,
+                    81.01
+                ],
+                &quot;inResponseTo&quot;: 4,
+                &quot;messageId&quot;: 3,
+                &quot;authorId&quot;: 2,
+                &quot;message&quot;: &quot; like product-y the plan is amazing&quot;
+            }
+        }
+    ],
+    &quot;uid&quot;: 2
+} ]
+</pre></div></div>
+
+<p>As we can see from the above query result, each group in the example query&#x2019;s output has an associated group variable value called <tt>msgs</tt> that appears in the <tt>SELECT *</tt>&#x2019;s result. This variable contains a collection of objects associated with the group; each of the group&#x2019;s <tt>message</tt> values appears in the <tt>msg</tt> field of the objects in the <tt>msgs</tt> collection.</p>
+<p>The group variable in the query language makes more complex, composable, nested subqueries over a group possible, which is important given the language&#x2019;s more complex data model (relative to SQL). As a simple example of this, as we really just want the messages associated with each user, we might wish to avoid the &#x201c;extra wrapping&#x201d; of each message as the <tt>msg</tt> field of an object. (That wrapping is useful in more complex cases, but is essentially just in the way here.) We can use a subquery in the <tt>SELECT</tt> clause to tunnel through the extra nesting and produce the desired result.</p></div>
+<div class="section">
+<h5><a name="Example"></a>Example</h5>
+
+<div>
+<div>
+<pre class="source">SELECT uid, (SELECT VALUE g.msg FROM g) AS msgs
+FROM GleambookMessages gbm
+GROUP BY gbm.authorId AS uid
+GROUP AS g(gbm as msg);
+</pre></div></div>
+
+<p>This variant of the example query returns:</p>
+
+<div>
+<div>
+<pre class="source">   [ {
+       &quot;msgs&quot;: [
+           {
+               &quot;senderLocation&quot;: [
+                   38.97,
+                   77.49
+               ],
+               &quot;inResponseTo&quot;: 1,
+               &quot;messageId&quot;: 11,
+               &quot;authorId&quot;: 1,
+               &quot;message&quot;: &quot; can't stand acast its plan is terrible&quot;
+           },
+           {
+               &quot;senderLocation&quot;: [
+                   41.66,
+                   80.87
+               ],
+               &quot;inResponseTo&quot;: 4,
+               &quot;messageId&quot;: 2,
+               &quot;authorId&quot;: 1,
+               &quot;message&quot;: &quot; dislike x-phone its touch-screen is horrible&quot;
+           },
+           {
+               &quot;senderLocation&quot;: [
+                   37.73,
+                   97.04
+               ],
+               &quot;inResponseTo&quot;: 2,
+               &quot;messageId&quot;: 4,
+               &quot;authorId&quot;: 1,
+               &quot;message&quot;: &quot; can't stand acast the network is horrible:(&quot;
+           },
+           {
+               &quot;senderLocation&quot;: [
+                   40.33,
+                   80.87
+               ],
+               &quot;inResponseTo&quot;: 11,
+               &quot;messageId&quot;: 8,
+               &quot;authorId&quot;: 1,
+               &quot;message&quot;: &quot; like ccast the 3G is awesome:)&quot;
+           },
+           {
+               &quot;senderLocation&quot;: [
+                   42.5,
+                   70.01
+               ],
+               &quot;inResponseTo&quot;: 12,
+               &quot;messageId&quot;: 10,
+               &quot;authorId&quot;: 1,
+               &quot;message&quot;: &quot; can't stand product-w the touch-screen is terrible&quot;
+           }
+       ],
+       &quot;uid&quot;: 1
+   }, {
+       &quot;msgs&quot;: [
+           {
+               &quot;senderLocation&quot;: [
+                   31.5,
+                   75.56
+               ],
+               &quot;inResponseTo&quot;: 1,
+               &quot;messageId&quot;: 6,
+               &quot;authorId&quot;: 2,
+               &quot;message&quot;: &quot; like product-z its platform is mind-blowing&quot;
+           },
+           {
+               &quot;senderLocation&quot;: [
+                   48.09,
+                   81.01
+               ],
+               &quot;inResponseTo&quot;: 4,
+               &quot;messageId&quot;: 3,
+               &quot;authorId&quot;: 2,
+               &quot;message&quot;: &quot; like product-y the plan is amazing&quot;
+           }
+       ],
+       &quot;uid&quot;: 2
+   } ]
+</pre></div></div>
+
+<p>The next example shows a more interesting case involving the use of a subquery in the <tt>SELECT</tt> list. Here the subquery further processes the groups. There is no renaming in the declaration of the group variable <tt>g</tt> such that <tt>g</tt> only has one field <tt>gbm</tt> which comes from the <tt>FROM</tt> clause.</p></div>
+<div class="section">
+<h5><a name="Example"></a>Example</h5>
+
+<div>
+<div>
+<pre class="source">SELECT uid,
+       (SELECT VALUE g.gbm
+        FROM g
+        WHERE g.gbm.message LIKE '% like%'
+        ORDER BY g.gbm.messageId
+        LIMIT 2) AS msgs
+FROM GleambookMessages gbm
+GROUP BY gbm.authorId AS uid
+GROUP AS g;
+</pre></div></div>
+
+<p>This example query returns:</p>
+
+<div>
+<div>
+<pre class="source">[ {
+    &quot;msgs&quot;: [
+        {
+            &quot;senderLocation&quot;: [
+                40.33,
+                80.87
+            ],
+            &quot;inResponseTo&quot;: 11,
+            &quot;messageId&quot;: 8,
+            &quot;authorId&quot;: 1,
+            &quot;message&quot;: &quot; like ccast the 3G is awesome:)&quot;
+        }
+    ],
+    &quot;uid&quot;: 1
+}, {
+    &quot;msgs&quot;: [
+        {
+            &quot;senderLocation&quot;: [
+                48.09,
+                81.01
+            ],
+            &quot;inResponseTo&quot;: 4,
+            &quot;messageId&quot;: 3,
+            &quot;authorId&quot;: 2,
+            &quot;message&quot;: &quot; like product-y the plan is amazing&quot;
+        },
+        {
+            &quot;senderLocation&quot;: [
+                31.5,
+                75.56
+            ],
+            &quot;inResponseTo&quot;: 1,
+            &quot;messageId&quot;: 6,
+            &quot;authorId&quot;: 2,
+            &quot;message&quot;: &quot; like product-z its platform is mind-blowing&quot;
+        }
+    ],
+    &quot;uid&quot;: 2
+} ]
+</pre></div></div>
+</div></div></div>
+<div class="section">
+<h3><a name="Implicit_Grouping_Key_Variables"></a><a name="Implicit_group_key_variables" id="Implicit_group_key_variables">Implicit Grouping Key Variables</a></h3>
+<p>In the query language syntax, providing named binding variables for <tt>GROUP BY</tt> key expressions is optional. If a grouping key is missing a user-provided binding variable, the underlying compiler will generate one. Automatic grouping key variable naming falls into three cases, much like the treatment of unnamed projections:</p>
+<ul>
+
+<li>If the grouping key expression is a variable reference expression, the generated variable gets the same name as the referred variable;</li>
+<li>If the grouping key expression is a field access expression, the generated variable gets the same name as the last identifier in the expression;</li>
+<li>For all other cases, the compiler generates a unique variable (but the user query is unable to refer to this generated variable).</li>
+</ul>
+<p>The next example illustrates a query that doesn&#x2019;t provide binding variables for its grouping key expressions.</p>
+<div class="section">
+<div class="section">
+<h5><a name="Example"></a>Example</h5>
+
+<div>
+<div>
+<pre class="source">SELECT authorId,
+       (SELECT VALUE g.gbm
+        FROM g
+        WHERE g.gbm.message LIKE '% like%'
+        ORDER BY g.gbm.messageId
+        LIMIT 2) AS msgs
+FROM GleambookMessages gbm
+GROUP BY gbm.authorId
+GROUP AS g;
+</pre></div></div>
+
+<p>This query returns:</p>
+
+<div>
+<div>
+<pre class="source">    [ {
+    &quot;msgs&quot;: [
+        {
+            &quot;senderLocation&quot;: [
+                40.33,
+                80.87
+            ],
+            &quot;inResponseTo&quot;: 11,
+            &quot;messageId&quot;: 8,
+            &quot;authorId&quot;: 1,
+            &quot;message&quot;: &quot; like ccast the 3G is awesome:)&quot;
+        }
+    ],
+    &quot;authorId&quot;: 1
+}, {
+    &quot;msgs&quot;: [
+        {
+            &quot;senderLocation&quot;: [
+                48.09,
+                81.01
+            ],
+            &quot;inResponseTo&quot;: 4,
+            &quot;messageId&quot;: 3,
+            &quot;authorId&quot;: 2,
+            &quot;message&quot;: &quot; like product-y the plan is amazing&quot;
+        },
+        {
+            &quot;senderLocation&quot;: [
+                31.5,
+                75.56
+            ],
+            &quot;inResponseTo&quot;: 1,
+            &quot;messageId&quot;: 6,
+            &quot;authorId&quot;: 2,
+            &quot;message&quot;: &quot; like product-z its platform is mind-blowing&quot;
+        }
+    ],
+    &quot;authorId&quot;: 2
+} ]
+</pre></div></div>
+
+<p>Based on the three variable generation rules, the generated variable for the grouping key expression <tt>message.authorId</tt> is <tt>authorId</tt> (which is how it is referred to in the example&#x2019;s <tt>SELECT</tt> clause).</p></div></div></div>
+<div class="section">
+<h3><a name="Implicit_Group_Variables"></a><a name="Implicit_group_variables" id="Implicit_group_variables">Implicit Group Variables</a></h3>
+<p>The group variable itself is also optional in the <tt>GROUP BY</tt> syntax. If a user&#x2019;s query does not declare the name and structure of the group variable using <tt>GROUP AS</tt>, the query compiler will generate a unique group variable whose fields include all of the binding variables defined in the <tt>FROM</tt> clause of the current enclosing <tt>SELECT</tt> statement. In this case the user&#x2019;s query will not be able to refer to the generated group variable, but is able to call SQL-92 aggregation functions as in SQL-92.</p></div>
+<div class="section">
+<h3><a name="Aggregation_Functions"></a><a name="Aggregation_functions" id="Aggregation_functions">Aggregation Functions</a></h3>
+<p>In the traditional SQL, which doesn&#x2019;t support nested data, grouping always also involves the use of aggregation to compute properties of the groups (for example, the average number of messages per user rather than the actual set of messages per user). Each aggregation function in the query language takes a collection (for example, the group of messages) as its input and produces a scalar value as its output. These aggregation functions, being truly functional in nature (unlike in SQL), can be used anywhere in a query where an expression is allowed. The following table catalogs the built-in aggregation functions of the query language and also indicates how each one handles <tt>NULL</tt>/<tt>MISSING</tt> values in the input collection or a completely empty input collection:</p>
+<table border="0" class="table table-striped">
+<thead>
+
+<tr class="a">
+<th> Function       </th>
+<th> NULL         </th>
+<th> MISSING      </th>
+<th> Empty Collection </th></tr>
+</thead><tbody>
+
+<tr class="b">
+<td> STRICT_COUNT   </td>
+<td> counted      </td>
+<td> counted      </td>
+<td> 0                </td></tr>
+<tr class="a">
+<td> STRICT_SUM     </td>
+<td> returns NULL </td>
+<td> returns NULL </td>
+<td> returns NULL     </td></tr>
+<tr class="b">
+<td> STRICT_MAX     </td>
+<td> returns NULL </td>
+<td> returns NULL </td>
+<td> returns NULL     </td></tr>
+<tr class="a">
+<td> STRICT_MIN     </td>
+<td> returns NULL </td>
+<td> returns NULL </td>
+<td> returns NULL     </td></tr>
+<tr class="b">
+<td> STRICT_AVG     </td>
+<td> returns NULL </td>
+<td> returns NULL </td>
+<td> returns NULL     </td></tr>
+<tr class="a">
+<td> STRICT_STDDEV_SAMP </td>
+<td> returns NULL </td>
+<td> returns NULL </td>
+<td> returns NULL </td></tr>
+<tr class="b">
+<td> STRICT_STDDEV_POP  </td>
+<td> returns NULL </td>
+<td> returns NULL </td>
+<td> returns NULL </td></tr>
+<tr class="a">
+<td> STRICT_VAR_SAMP    </td>
+<td> returns NULL </td>
+<td> returns NULL </td>
+<td> returns NULL </td></tr>
+<tr class="b">
+<td> STRICT_VAR_POP     </td>
+<td> returns NULL </td>
+<td> returns NULL </td>
+<td> returns NULL </td></tr>
+<tr class="a">
+<td> STRICT_SKEWNESS    </td>
+<td> returns NULL </td>
+<td> returns NULL </td>
+<td> returns NULL </td></tr>
+<tr class="b">
+<td> STRICT_KURTOSIS    </td>
+<td> returns NULL </td>
+<td> returns NULL </td>
+<td> returns NULL </td></tr>
+<tr class="a">
+<td> ARRAY_COUNT    </td>
+<td> not counted  </td>
+<td> not counted  </td>
+<td> 0                </td></tr>
+<tr class="b">
+<td> ARRAY_SUM      </td>
+<td> ignores NULL </td>
+<td> ignores NULL </td>
+<td> returns NULL     </td></tr>
+<tr class="a">
+<td> ARRAY_MAX      </td>
+<td> ignores NULL </td>
+<td> ignores NULL </td>
+<td> returns NULL     </td></tr>
+<tr class="b">
+<td> ARRAY_MIN      </td>
+<td> ignores NULL </td>
+<td> ignores NULL </td>
+<td> returns NULL     </td></tr>
+<tr class="a">
+<td> ARRAY_AVG      </td>
+<td> ignores NULL </td>
+<td> ignores NULL </td>
+<td> returns NULL     </td></tr>
+<tr class="b">
+<td> ARRAY_STDDEV_SAMP  </td>
+<td> ignores NULL </td>
+<td> ignores NULL </td>
+<td> returns NULL </td></tr>
+<tr class="a">
+<td> ARRAY_STDDEV_POP   </td>
+<td> ignores NULL </td>
+<td> ignores NULL </td>
+<td> returns NULL </td></tr>
+<tr class="b">
+<td> ARRAY_VAR_SAMP     </td>
+<td> ignores NULL </td>
+<td> ignores NULL </td>
+<td> returns NULL </td></tr>
+<tr class="a">
+<td> ARRAY_VAR_POP      </td>
+<td> ignores NULL </td>
+<td> ignores NULL </td>
+<td> returns NULL </td></tr>
+<tr class="b">
+<td> ARRAY_SKEWNESS     </td>
+<td> ignores NULL </td>
+<td> ignores NULL </td>
+<td> returns NULL </td></tr>
+<tr class="a">
+<td> ARRAY_KURTOSIS     </td>
+<td> ignores NULL </td>
+<td> ignores NULL </td>
+<td> returns NULL </td></tr>
+</tbody>
+</table>
+<p>Notice that the query language offers two versions for each of the aggregate functions listed above. For each function, the STRICT version handles <tt>UNKNOWN</tt> values in a semantically strict fashion, where unknown values in the input result in unknown values in the output; and the ARRAY version handles them in the ad hoc &#x201c;just ignore the unknown values&#x201d; fashion that the SQL standard chose to adopt.</p>
+<div class="section">
+<div class="section">
+<h5><a name="Example"></a>Example</h5>
+
+<div>
+<div>
+<pre class="source">ARRAY_AVG(
+    (
+      SELECT VALUE ARRAY_COUNT(friendIds) FROM GleambookUsers
+    )
+);
+</pre></div></div>
+
+<p>This example returns:</p>
+
+<div>
+<div>
+<pre class="source">3.3333333333333335
+</pre></div></div>
+</div>
+<div class="section">
+<h5><a name="Example"></a>Example</h5>
+
+<div>
+<div>
+<pre class="source">SELECT uid AS uid, ARRAY_COUNT(grp) AS msgCnt
+FROM GleambookMessages message
+GROUP BY message.authorId AS uid
+GROUP AS grp(message AS msg);
+</pre></div></div>
+
+<p>This query returns:</p>
+
+<div>
+<div>
+<pre class="source">[ {
+    &quot;uid&quot;: 1,
+    &quot;msgCnt&quot;: 5
+}, {
+    &quot;uid&quot;: 2,
+    &quot;msgCnt&quot;: 2
+} ]
+</pre></div></div>
+
+<p>Notice how the query forms groups where each group involves a message author and their messages. (SQL cannot do this because the grouped intermediate result is non-1NF in nature.) The query then uses the collection aggregate function ARRAY_COUNT to get the cardinality of each group of messages.</p>
+<p>Each aggregation function in the query language supports the DISTINCT modifier that removes duplicate values from the input collection.</p></div>
+<div class="section">
+<h5><a name="Example"></a>Example</h5>
+
+<div>
+<div>
+<pre class="source">ARRAY_SUM(DISTINCT [1, 1, 2, 2, 3])
+</pre></div></div>
+
+<p>This query returns:</p>
+
+<div>
+<div>
+<pre class="source">6
+</pre></div></div>
+</div></div></div>
+<div class="section">
+<h3><a name="SQL-92_Aggregation_Functions"></a><a name="SQL-92_aggregation_functions" id="SQL-92_aggregation_functions">SQL-92 Aggregation Functions</a></h3>
+<p>For compatibility with the traditional SQL aggregation functions, the query language also offers SQL-92&#x2019;s aggregation function symbols (<tt>COUNT</tt>, <tt>SUM</tt>, <tt>MAX</tt>, <tt>MIN</tt>, <tt>AVG</tt>, <tt>ARRAY_AGG</tt>, <tt>STDDEV_SAMP</tt>, <tt>STDDEV_POP</tt>, <tt>VAR_SAMP</tt>, <tt>VAR_POP</tt>) as supported syntactic sugar. The query compiler rewrites queries that utilize these function symbols into queries that only use the collection aggregate functions of the query language. The following example uses the SQL-92 syntax approach to compute a result that is identical to that of the more explicit example above:</p>
+<div class="section">
+<div class="section">
+<h5><a name="Example"></a>Example</h5>
+
+<div>
+<div>
+<pre class="source">SELECT uid, COUNT(*) AS msgCnt
+FROM GleambookMessages msg
+GROUP BY msg.authorId AS uid;
+</pre></div></div>
+
+<p>It is important to realize that <tt>COUNT</tt> is actually <b>not</b> a built-in aggregation function. Rather, the <tt>COUNT</tt> query above is using a special &#x201c;sugared&#x201d; function symbol that the query compiler will rewrite as follows:</p>
+
+<div>
+<div>
+<pre class="source">SELECT uid AS uid, ARRAY_COUNT( (SELECT VALUE 1 FROM `$1` AS g) ) AS msgCnt
+FROM GleambookMessages msg
+GROUP BY msg.authorId AS uid
+GROUP AS `$1`(msg AS msg);
+</pre></div></div>
+
+<p>The same sort of rewritings apply to the function symbols <tt>SUM</tt>, <tt>MAX</tt>, <tt>MIN</tt>, <tt>AVG</tt>, <tt>ARRAY_AGG</tt>,<tt>STDDEV_SAMP</tt>, <tt>STDDEV_POP</tt>, <tt>VAR_SAMP</tt>, and <tt>VAR_POP</tt>. In contrast to the collection aggregate functions of the query language, these special SQL-92 function symbols can only be used in the same way they are in standard SQL (i.e., with the same restrictions).</p>
+<p>The DISTINCT modifier is also supported for these aggregate functions.</p>
+<p>The following table shows the SQL-92 functions supported by the query language, their aliases where available, and their corresponding built-in functions.</p>
+<table border="0" class="table table-striped">
+<thead>
+
+<tr class="a">
+<th> SQL-92 Function </th>
+<th> Aliases                 </th>
+<th> Corresponding Built-in Function </th></tr>
+</thead><tbody>
+
+<tr class="b">
+<td> COUNT           </td>
+<td>                         </td>
+<td> ARRAY_COUNT                     </td></tr>
+<tr class="a">
+<td> SUM             </td>
+<td>                         </td>
+<td> ARRAY_SUM                       </td></tr>
+<tr class="b">
+<td> MAX             </td>
+<td>                         </td>
+<td> ARRAY_MAX                       </td></tr>
+<tr class="a">
+<td> MIN             </td>
+<td>                         </td>
+<td> ARRAY_MIN                       </td></tr>
+<tr class="b">
+<td> AVG             </td>
+<td>                         </td>
+<td> ARRAY_AVG                       </td></tr>
+<tr class="a">
+<td> ARRAY_AGG       </td>
+<td>                         </td>
+<td> (none)                          </td></tr>
+<tr class="b">
+<td> STDDEV_SAMP     </td>
+<td> STDDEV                  </td>
+<td> ARRAY_STDDEV_SAMP               </td></tr>
+<tr class="a">
+<td> STDDEV_POP      </td>
+<td>                         </td>
+<td> ARRAY_STDDEV_POP                </td></tr>
+<tr class="b">
+<td> VAR_SAMP        </td>
+<td> VARIANCE, VARIANCE_SAMP </td>
+<td> ARRAY_VAR_SAMP                  </td></tr>
+<tr class="a">
+<td> VAR_POP         </td>
+<td> VARIANCE_POP            </td>
+<td> ARRAY_VAR_POP                   </td></tr>
+</tbody>
+</table>
+<p>Note that the <tt>ARRAY_AGG</tt> function symbol is rewritten simply to return the result of the generated subquery, without applying any built-in function.</p>
+<p>SQL aggregate function calls optionally support a FILTER subclause.</p></div>
+<div class="section">
+<h5><a name="Example"></a>Example</h5>
+
+<div>
+<div>
+<pre class="source">SELECT uid, COUNT(*) FILTER (WHERE msg.message LIKE &quot;%awesome%&quot;) AS msgCnt
+FROM GleambookMessages msg
+GROUP BY msg.authorId AS uid;
+</pre></div></div>
+
+<p>The query compiler rewrites this query to use the built-in aggregate as follows:</p>
+
+<div>
+<div>
+<pre class="source">SELECT uid AS uid, ARRAY_COUNT( (SELECT VALUE 1 FROM `$1` AS g WHERE g.msg.message LIKE &quot;%awesome%&quot;) ) AS msgCnt
+FROM GleambookMessages msg
+GROUP BY msg.authorId AS uid
+GROUP AS `$1`(msg AS msg);
+</pre></div></div>
+
+<p>Note that the FILTER subclause is not supported for built-in aggregate function calls.</p></div></div></div>
+<div class="section">
+<h3><a name="SQL-92_Compliant_GROUP_BY_Aggregations"></a><a name="SQL-92_compliant_gby" id="SQL-92_compliant_gby">SQL-92 Compliant GROUP BY Aggregations</a></h3>
+<p>The query language provides full support for SQL-92 <tt>GROUP BY</tt> aggregation queries. The following query is such an example:</p>
+<div class="section">
+<div class="section">
+<h5><a name="Example"></a>Example</h5>
+
+<div>
+<div>
+<pre class="source">SELECT msg.authorId, COUNT(*)
+FROM GleambookMessages msg
+GROUP BY msg.authorId;
+</pre></div></div>
+
+<p>This query outputs:</p>
+
+<div>
+<div>
+<pre class="source">[ {
+    &quot;authorId&quot;: 1,
+    &quot;$1&quot;: 5
+}, {
+    &quot;authorId&quot;: 2,
+    &quot;$1&quot;: 2
+} ]
+</pre></div></div>
+
+<p>In principle, a <tt>msg</tt> reference in the query&#x2019;s <tt>SELECT</tt> clause would be &#x201c;sugarized&#x201d; as a collection (as described in <a href="#Implicit_group_variables">Implicit Group Variables</a>). However, since the SELECT expression <tt>msg.authorId</tt> is syntactically identical to a GROUP BY key expression, it will be internally replaced by the generated group key variable. The following is the equivalent rewritten query that will be generated by the compiler for the query above:</p>
+
+<div>
+<div>
+<pre class="source">SELECT authorId AS authorId, ARRAY_COUNT( (SELECT g.msg FROM `$1` AS g) )
+FROM GleambookMessages msg
+GROUP BY msg.authorId AS authorId
+GROUP AS `$1`(msg AS msg);
+</pre></div></div>
+</div></div></div>
+<div class="section">
+<h3><a name="Column_Aliases"></a><a name="Column_aliases" id="Column_aliases">Column Aliases</a></h3>
+<p>The query language also allows column aliases to be used as <tt>ORDER BY</tt> keys.</p>
+<div class="section">
+<div class="section">
+<h5><a name="Example"></a>Example</h5>
+
+<div>
+<div>
+<pre class="source">SELECT msg.authorId AS aid, COUNT(*)
+FROM GleambookMessages msg
+GROUP BY msg.authorId;
+ORDER BY aid;
+</pre></div></div>
+
+<p>This query returns:</p>
+
+<div>
+<div>
+<pre class="source">[ {
+    &quot;$1&quot;: 5,
+    &quot;aid&quot;: 1
+}, {
+    &quot;$1&quot;: 2,
+    &quot;aid&quot;: 2
+} ]
+</pre></div></div>
+</div></div></div></div>
+<div class="section">
+<h2><a name="WHERE_Clauses_and_HAVING_Clauses"></a><a name="Where_having_clauses" id="Where_having_clauses">WHERE Clauses and HAVING Clauses</a></h2>
+<p>Both <tt>WHERE</tt> clauses and <tt>HAVING</tt> clauses are used to filter input data based on a condition expression. Only tuples for which the condition expression evaluates to <tt>TRUE</tt> are propagated. Note that if the condition expression evaluates to <tt>NULL</tt> or <tt>MISSING</tt> the input tuple will be discarded.</p></div>
+<div class="section">
+<h2><a name="ORDER_BY_Clauses"></a><a name="Order_By_clauses" id="Order_By_clauses">ORDER BY Clauses</a></h2>
+<p>The <tt>ORDER BY</tt> clause is used to globally sort data in either ascending order (i.e., <tt>ASC</tt>) or descending order (i.e., <tt>DESC</tt>). During ordering, <tt>MISSING</tt> and <tt>NULL</tt> are treated as being smaller than any other value if they are encountered in the ordering key(s). <tt>MISSING</tt> is treated as smaller than <tt>NULL</tt> if both occur in the data being sorted. The ordering of values of a given type is consistent with its type&#x2019;s &lt;= ordering; the ordering of values across types is implementation-defined but stable. The following example returns all <tt>GleambookUsers</tt> in descending order by their number of friends.</p>
+<div class="section">
+<div class="section">
+<div class="section">
+<h5><a name="Example"></a>Example</h5>
+
+<div>
+<div>
+<pre class="source">  SELECT VALUE user
+  FROM GleambookUsers AS user
+  ORDER BY ARRAY_COUNT(user.friendIds) DESC;
+</pre></div></div>
+
+<p>This query returns:</p>
+
+<div>
+<div>
+<pre class="source">  [ {
+      &quot;userSince&quot;: &quot;2012-08-20T10:10:00.000Z&quot;,
+      &quot;friendIds&quot;: [
+          2,
+          3,
+          6,
+          10
+      ],
+      &quot;gender&quot;: &quot;F&quot;,
+      &quot;name&quot;: &quot;MargaritaStoddard&quot;,
+      &quot;nickname&quot;: &quot;Mags&quot;,
+      &quot;alias&quot;: &quot;Margarita&quot;,
+      &quot;id&quot;: 1,
+      &quot;employment&quot;: [
+          {
+              &quot;organizationName&quot;: &quot;Codetechno&quot;,
+              &quot;start-date&quot;: &quot;2006-08-06&quot;
+          },
+          {
+              &quot;end-date&quot;: &quot;2010-01-26&quot;,
+              &quot;organizationName&quot;: &quot;geomedia&quot;,
+              &quot;start-date&quot;: &quot;2010-06-17&quot;
+          }
+      ]
+  }, {
+      &quot;userSince&quot;: &quot;2012-07-10T10:10:00.000Z&quot;,
+      &quot;friendIds&quot;: [
+          1,
+          5,
+          8,
+          9
+      ],
+      &quot;name&quot;: &quot;EmoryUnk&quot;,
+      &quot;alias&quot;: &quot;Emory&quot;,
+      &quot;id&quot;: 3,
+      &quot;employment&quot;: [
+          {
+              &quot;organizationName&quot;: &quot;geomedia&quot;,
+              &quot;endDate&quot;: &quot;2010-01-26&quot;,
+              &quot;startDate&quot;: &quot;2010-06-17&quot;
+          }
+      ]
+  }, {
+      &quot;userSince&quot;: &quot;2011-01-22T10:10:00.000Z&quot;,
+      &quot;friendIds&quot;: [
+          1,
+          4
+      ],
+      &quot;name&quot;: &quot;IsbelDull&quot;,
+      &quot;nickname&quot;: &quot;Izzy&quot;,
+      &quot;alias&quot;: &quot;Isbel&quot;,
+      &quot;id&quot;: 2,
+      &quot;employment&quot;: [
+          {
+              &quot;organizationName&quot;: &quot;Hexviafind&quot;,
+              &quot;startDate&quot;: &quot;2010-04-27&quot;
+          }
+      ]
+  } ]
+</pre></div></div>
+</div></div></div></div>
+<div class="section">
+<h2><a name="LIMIT_Clauses"></a><a name="Limit_clauses" id="Limit_clauses">LIMIT Clauses</a></h2>
+<p>The <tt>LIMIT</tt> clause is used to limit the result set to a specified constant size. The use of the <tt>LIMIT</tt> clause is illustrated in the next example.</p>
+<div class="section">
+<div class="section">
+<div class="section">
+<h5><a name="Example"></a>Example</h5>
+
+<div>
+<div>
+<pre class="source">  SELECT VALUE user
+  FROM GleambookUsers AS user
+  ORDER BY len(user.friendIds) DESC
+  LIMIT 1;
+</pre></div></div>
+
+<p>This query returns:</p>
+
+<div>
+<div>
+<pre class="source">  [ {
+      &quot;userSince&quot;: &quot;2012-08-20T10:10:00.000Z&quot;,
+      &quot;friendIds&quot;: [
+          2,
+          3,
+          6,
+          10
+      ],
+      &quot;gender&quot;: &quot;F&quot;,
+      &quot;name&quot;: &quot;MargaritaStoddard&quot;,
+      &quot;nickname&quot;: &quot;Mags&quot;,
+      &quot;alias&quot;: &quot;Margarita&quot;,
+      &quot;id&quot;: 1,
+      &quot;employment&quot;: [
+          {
+              &quot;organizationName&quot;: &quot;Codetechno&quot;,
+              &quot;start-date&quot;: &quot;2006-08-06&quot;
+          },
+          {
+              &quot;end-date&quot;: &quot;2010-01-26&quot;,
+              &quot;organizationName&quot;: &quot;geomedia&quot;,
+              &quot;start-date&quot;: &quot;2010-06-17&quot;
+          }
+      ]
+  } ]
+</pre></div></div>
+</div></div></div></div>
+<div class="section">
+<h2><a name="WITH_Clauses"></a><a name="With_clauses" id="With_clauses">WITH Clauses</a></h2>
+<p>As in standard SQL, <tt>WITH</tt> clauses are available to improve the modularity of a query. The next query shows an example.</p>
+<div class="section">
+<div class="section">
+<div class="section">
+<h5><a name="Example"></a>Example</h5>
+
+<div>
+<div>
+<pre class="source">WITH avgFriendCount AS (
+  SELECT VALUE AVG(ARRAY_COUNT(user.friendIds))
+  FROM GleambookUsers AS user
+)[0]
+SELECT VALUE user
+FROM GleambookUsers user
+WHERE ARRAY_COUNT(user.friendIds) &gt; avgFriendCount;
+</pre></div></div>
+
+<p>This query returns:</p>
+
+<div>
+<div>
+<pre class="source">[ {
+    &quot;userSince&quot;: &quot;2012-08-20T10:10:00.000Z&quot;,
+    &quot;friendIds&quot;: [
+        2,
+        3,
+        6,
+        10
+    ],
+    &quot;gender&quot;: &quot;F&quot;,
+    &quot;name&quot;: &quot;MargaritaStoddard&quot;,
+    &quot;nickname&quot;: &quot;Mags&quot;,
+    &quot;alias&quot;: &quot;Margarita&quot;,
+    &quot;id&quot;: 1,
+    &quot;employment&quot;: [
+        {
+            &quot;organizationName&quot;: &quot;Codetechno&quot;,
+            &quot;start-date&quot;: &quot;2006-08-06&quot;
+        },
+        {
+            &quot;end-date&quot;: &quot;2010-01-26&quot;,
+            &quot;organizationName&quot;: &quot;geomedia&quot;,
+            &quot;start-date&quot;: &quot;2010-06-17&quot;
+        }
+    ]
+}, {
+    &quot;userSince&quot;: &quot;2012-07-10T10:10:00.000Z&quot;,
+    &quot;friendIds&quot;: [
+        1,
+        5,
+        8,
+        9
+    ],
+    &quot;name&quot;: &quot;EmoryUnk&quot;,
+    &quot;alias&quot;: &quot;Emory&quot;,
+    &quot;id&quot;: 3,
+    &quot;employment&quot;: [
+        {
+            &quot;organizationName&quot;: &quot;geomedia&quot;,
+            &quot;endDate&quot;: &quot;2010-01-26&quot;,
+            &quot;startDate&quot;: &quot;2010-06-17&quot;
+        }
+    ]
+} ]
+</pre></div></div>
+
+<p>The query is equivalent to the following, more complex, inlined form of the query:</p>
+
+<div>
+<div>
+<pre class="source">SELECT *
+FROM GleambookUsers user
+WHERE ARRAY_COUNT(user.friendIds) &gt;
+    ( SELECT VALUE AVG(ARRAY_COUNT(user.friendIds))
+      FROM GleambookUsers AS user
+    ) [0];
+</pre></div></div>
+
+<p>WITH can be particularly useful when a value needs to be used several times in a query.</p>
+<p>Before proceeding further, notice that both the WITH query and its equivalent inlined variant include the syntax &#x201c;[0]&#x201d; &#x2013; this is due to a noteworthy difference between the query language and SQL-92. In SQL-92, whenever a scalar value is expected and it is being produced by a query expression, the SQL-92 query processor will evaluate the expression, check that there is only one row and column in the result at runtime, and then coerce the one-row/one-column tabular result into a scalar value. A JSON query language, being designed to deal with nested data and schema-less data, should not do this. Collection-valued data is perfectly legal in most contexts, and its data is schema-less, so the query processor rarely knows exactly what to expect where and such automatic conversion would often not be desirable. Thus, in the queries above, the use of &#x201c;[0]&#x201d; extracts the first (i.e., 0th) element of an array-valued query expression&#x2019;s result; this is needed above, even though the result is an array of one element, to extract the only element in the singleton array and obtain the desired scalar for the comparison.</p></div></div></div></div>
+<div class="section">
+<h2><a name="LET_Clauses"></a><a name="Let_clauses" id="Let_clauses">LET Clauses</a></h2>
+<p>Similar to <tt>WITH</tt> clauses, <tt>LET</tt> clauses can be useful when a (complex) expression is used several times within a query, allowing it to be written once to make the query more concise. The next query shows an example.</p>
+<div class="section">
+<div class="section">
+<div class="section">
+<h5><a name="Example"></a>Example</h5>
+
+<div>
+<div>
+<pre class="source">SELECT u.name AS uname, messages AS messages
+FROM GleambookUsers u
+LET messages = (SELECT VALUE m
+                FROM GleambookMessages m
+                WHERE m.authorId = u.id)
+WHERE EXISTS messages;
+</pre></div></div>
+
+<p>This query lists <tt>GleambookUsers</tt> that have posted <tt>GleambookMessages</tt> and shows all authored messages for each listed user. It returns:</p>
+
+<div>
+<div>
+<pre class="source">[ {
+    &quot;uname&quot;: &quot;MargaritaStoddard&quot;,
+    &quot;messages&quot;: [
+        {
+            &quot;senderLocation&quot;: [
+                38.97,
+                77.49
+            ],
+            &quot;inResponseTo&quot;: 1,
+            &quot;messageId&quot;: 11,
+            &quot;authorId&quot;: 1,
+            &quot;message&quot;: &quot; can't stand acast its plan is terrible&quot;
+        },
+        {
+            &quot;senderLocation&quot;: [
+                41.66,
+                80.87
+            ],
+            &quot;inResponseTo&quot;: 4,
+            &quot;messageId&quot;: 2,
+            &quot;authorId&quot;: 1,
+            &quot;message&quot;: &quot; dislike x-phone its touch-screen is horrible&quot;
+        },
+        {
+            &quot;senderLocation&quot;: [
+                37.73,
+                97.04
+            ],
+            &quot;inResponseTo&quot;: 2,
+            &quot;messageId&quot;: 4,
+            &quot;authorId&quot;: 1,
+            &quot;message&quot;: &quot; can't stand acast the network is horrible:(&quot;
+        },
+        {
+            &quot;senderLocation&quot;: [
+                40.33,
+                80.87
+            ],
+            &quot;inResponseTo&quot;: 11,
+            &quot;messageId&quot;: 8,
+            &quot;authorId&quot;: 1,
+            &quot;message&quot;: &quot; like ccast the 3G is awesome:)&quot;
+        },
+        {
+            &quot;senderLocation&quot;: [
+                42.5,
+                70.01
+            ],
+            &quot;inResponseTo&quot;: 12,
+            &quot;messageId&quot;: 10,
+            &quot;authorId&quot;: 1,
+            &quot;message&quot;: &quot; can't stand product-w the touch-screen is terrible&quot;
+        }
+    ]
+}, {
+    &quot;uname&quot;: &quot;IsbelDull&quot;,
+    &quot;messages&quot;: [
+        {
+            &quot;senderLocation&quot;: [
+                31.5,
+                75.56
+            ],
+            &quot;inResponseTo&quot;: 1,
+            &quot;messageId&quot;: 6,
+            &quot;authorId&quot;: 2,
+            &quot;message&quot;: &quot; like product-z its platform is mind-blowing&quot;
+        },
+        {
+            &quot;senderLocation&quot;: [
+                48.09,
+                81.01
+            ],
+            &quot;inResponseTo&quot;: 4,
+            &quot;messageId&quot;: 3,
+            &quot;authorId&quot;: 2,
+            &quot;message&quot;: &quot; like product-y the plan is amazing&quot;
+        }
+    ]
+} ]
+</pre></div></div>
+
+<p>This query is equivalent to the following query that does not use the <tt>LET</tt> clause:</p>
+
+<div>
+<div>
+<pre class="source">SELECT u.name AS uname, ( SELECT VALUE m
+                          FROM GleambookMessages m
+                          WHERE m.authorId = u.id
+                        ) AS messages
+FROM GleambookUsers u
+WHERE EXISTS ( SELECT VALUE m
+               FROM GleambookMessages m
+               WHERE m.authorId = u.id
+             );
+</pre></div></div>
+</div></div></div></div>
+<div class="section">
+<h2><a name="UNION_ALL"></a><a name="Union_all" id="Union_all">UNION ALL</a></h2>
+<p>UNION ALL can be used to combine two input arrays or multisets into one. As in SQL, there is no ordering guarantee on the contents of the output stream. However, unlike SQL, the query language does not constrain what the data looks like on the input streams; in particular, it allows heterogeneity on the input and output streams. A type error will be raised if one of the inputs is not a collection. The following odd but legal query is an example:</p>
+<div class="section">
+<div class="section">
+<div class="section">
+<h5><a name="Example"></a>Example</h5>
+
+<div>
+<div>
+<pre class="source">SELECT u.name AS uname
+FROM GleambookUsers u
+WHERE u.id = 2
+  UNION ALL
+SELECT VALUE m.message
+FROM GleambookMessages m
+WHERE authorId=2;
+</pre></div></div>
+
+<p>This query returns:</p>
+
+<div>
+<div>
+<pre class="source">[
+  &quot; like product-z its platform is mind-blowing&quot;
+  , {
+    &quot;uname&quot;: &quot;IsbelDull&quot;
+}, &quot; like product-y the plan is amazing&quot;
+ ]
+</pre></div></div>
+</div></div></div></div>
+<div class="section">
+<h2><a name="OVER_Clauses"></a><a name="Over_clauses" id="Over_clauses">OVER Clauses</a></h2>
+<p>All window functions must have an OVER clause to define the window partitions, the order of tuples within those partitions, and the extent of the window frame. Some window functions take additional window options, which are specified by modifiers before the OVER clause.</p>
+<p>The query language has a dedicated set of window functions. Aggregate functions can also be used as window functions, when they are used with an OVER clause.</p>
+<div class="section">
+<h3><a name="Window_Function_Call"></a><a name="Window_function_call" id="Window_function_call">Window Function Call</a></h3>
+
+<div>
+<div>
+<pre class="source">WindowFunctionCall ::= WindowFunctionType &quot;(&quot; WindowFunctionArguments &quot;)&quot;
+(WindowFunctionOptions)? &lt;OVER&gt; (Variable &lt;AS&gt;)? &quot;(&quot; WindowDefinition &quot;)&quot;
+</pre></div></div>
+
+<div class="section">
+<h4><a name="Window_Function_Type"></a><a name="Window_function_type" id="Window_function_type">Window Function Type</a></h4>
+
+<div>
+<div>
+<pre class="source">WindowFunctionType ::= AggregateFunction | WindowFunction
+</pre></div></div>
+
+<p>Refer to the <a href="builtins.html#AggregateFunctions">Aggregate Functions</a> section for a list of aggregate functions.</p>
+<p>Refer to the <a href="builtins.html#WindowFunctions">Window Functions</a> section for a list of window functions.</p></div>
+<div class="section">
+<h4><a name="Window_Function_Arguments"></a><a name="Window_function_arguments" id="Window_function_arguments">Window Function Arguments</a></h4>
+
+<div>
+<div>
+<pre class="source">WindowFunctionArguments ::= ( (&lt;DISTINCT&gt;)? Expression |
+(Expression (&quot;,&quot; Expression (&quot;,&quot; Expression)? )? )? )
+</pre></div></div>
+
+<p>Refer to the <a href="builtins.html#AggregateFunctions">Aggregate Functions</a> section or the <a href="builtins.html#WindowFunctions">Window Functions</a> section for details of the arguments for individual functions.</p></div></div>
+<div class="section">
+<h3><a name="Window_Function_Options"></a><a name="Window_function_options" id="Window_function_options">Window Function Options</a></h3>
+
+<div>
+<div>
+<pre class="source">WindowFunctionOptions ::= (NthValFrom)? (NullsTreatment)?
+</pre></div></div>
+
+<p>Window function options cannot be used with <a href="builtins.html#AggregateFunctions">aggregate functions</a>.</p>
+<p>Window function options can only be used with some <a href="builtins.html#WindowFunctions">window functions</a>, as described below.</p>
+<div class="section">
+<h4><a name="Nth_Val_From"></a><a name="Nth_val_from" id="Nth_val_from">Nth Val From</a></h4>
+
+<div>
+<div>
+<pre class="source">NthValFrom ::= &lt;FROM&gt; ( &lt;FIRST&gt; | &lt;LAST&gt; )
+</pre></div></div>
+
+<p>The <b>nth val from</b> modifier determines whether the computation begins at the first or last tuple in the window.</p>
+<p>This modifier can only be used with the <tt>nth_value()</tt> function.</p>
+<p>This modifier is optional. If omitted, the default setting is <tt>FROM FIRST</tt>.</p></div>
+<div class="section">
+<h4><a name="Nulls_Treatment"></a><a name="Nulls_treatment" id="Nulls_treatment">Nulls Treatment</a></h4>
+
+<div>
+<div>
+<pre class="source">NullsTreatment ::= ( &lt;RESPECT&gt; | &lt;IGNORE&gt; ) &lt;NULLS&gt;
+</pre></div></div>
+
+<p>The <b>nulls treatment</b> modifier determines whether NULL values are included in the computation, or ignored. MISSING values are treated the same way as NULL values.</p>
+<p>This modifier can only be used with the <tt>first_value()</tt>, <tt>last_value()</tt>, <tt>nth_value()</tt>, <tt>lag()</tt>, and <tt>lead()</tt> functions.</p>
+<p>This modifier is optional. If omitted, the default setting is <tt>RESPECT NULLS</tt>.</p></div></div>
+<div class="section">
+<h3><a name="Window_Frame_Variable"></a><a name="Window_frame_variable" id="Window_frame_variable">Window Frame Variable</a></h3>
+<p>The AS keyword enables you to specify an alias for the window frame contents. It introduces a variable which will be bound to the contents of the frame. When using a built-in <a href="builtins.html#AggregateFunctions">aggregate function</a> as a window function, the function&#x2019;s argument must be a subquery which refers to this alias, for example:</p>
+
+<div>
+<div>
+<pre class="source">SELECT ARRAY_COUNT(DISTINCT (FROM alias SELECT VALUE alias.src.field))
+OVER alias AS (PARTITION BY &#x2026; ORDER BY &#x2026;)
+FROM source AS src
+</pre></div></div>
+
+<p>The alias is not necessary when using a <a href="builtins.html#WindowFunctions">window function</a>, or when using a standard SQL aggregate function with the OVER clause.</p>
+<div class="section">
+<h4><a name="Standard_SQL_Aggregate_Functions_with_the_OVER_Clause"></a><a name="SQL-92_over_clause" id="SQL-92_over_clause">Standard SQL Aggregate Functions with the OVER Clause</a></h4>
+<p>A standard SQL aggregate function with an OVER clause is rewritten by the query compiler using a built-in aggregate function over a frame variable. For example, the following query with the <tt>sum()</tt> function:</p>
+
+<div>
+<div>
+<pre class="source">SELECT SUM(field) OVER (PARTITION BY &#x2026; ORDER BY &#x2026;)
+FROM source AS src
+</pre></div></div>
+
+<p>Is rewritten as the following query using the <tt>array_sum()</tt> function:</p>
+
+<div>
+<div>
+<pre class="source">SELECT ARRAY_SUM( (SELECT VALUE alias.src.field FROM alias) )
+  OVER alias AS (PARTITION BY &#x2026; ORDER BY &#x2026;)
+FROM source AS src
+</pre></div></div>
+
+<p>This is similar to the way that standard SQL aggregate functions are rewritten as built-in aggregate functions in the presence of the GROUP BY clause.</p></div></div>
+<div class="section">
+<h3><a name="Window_Definition"></a><a name="Window_definition" id="Window_definition">Window Definition</a></h3>
+
+<div>
+<div>
+<pre class="source">WindowDefinition ::= (WindowPartitionClause)? (WindowOrderClause
+(WindowFrameClause (WindowFrameExclusion)? )? )?
+</pre></div></div>
+
+<p>The <b>window definition</b> specifies the partitioning, ordering, and framing for window functions.</p>
+<div class="section">
+<h4><a name="Window_Partition_Clause"></a><a name="Window_partition_clause" id="Window_partition_clause">Window Partition Clause</a></h4>
+
+<div>
+<div>
+<pre class="source">WindowPartitionClause ::= &lt;PARTITION&gt; &lt;BY&gt; Expression (&quot;,&quot; Expression)*
+</pre></div></div>
+
+<p>The <b>window partition clause</b> divides the tuples into logical partitions using one or more expressions.</p>
+<p>This clause may be used with any <a href="builtins.html#WindowFunctions">window function</a>, or any <a href="builtins.html#AggregateFunctions">aggregate function</a> used as a window function.</p>
+<p>This clause is optional. If omitted, all tuples are united in a single partition.</p></div>
+<div class="section">
+<h4><a name="Window_Order_Clause"></a><a name="Window_order_clause" id="Window_order_clause">Window Order Clause</a></h4>
+
+<div>
+<div>
+<pre class="source">WindowOrderClause ::= &lt;ORDER&gt; &lt;BY&gt; OrderingTerm (&quot;,&quot; OrderingTerm)*
+</pre></div></div>
+
+<p>The <b>window order clause</b> determines how tuples are ordered within each partition. The window function works on tuples in the order specified by this clause.</p>
+<p>This clause may be used with any <a href="builtins.html#WindowFunctions">window function</a>, or any <a href="builtins.html#AggregateFunctions">aggregate function</a> used as a window function.</p>
+<p>This clause is optional. If omitted, all tuples are considered peers, i.e. their order is tied. When tuples in the window partition are tied, each window function behaves differently.</p>
+<ul>
+
+<li>
+
+<p>The <tt>row_number()</tt> function returns a distinct number for each tuple. If tuples are tied, the results may be unpredictable.</p>
+</li>
+<li>
+
+<p>The <tt>rank()</tt>, <tt>dense_rank()</tt>, <tt>percent_rank()</tt>, and <tt>cume_dist()</tt> functions return the same result for each tuple.</p>
+</li>
+<li>
+
+<p>For other functions, if the <a href="#Window_frame_clause">window frame</a> is defined by <tt>ROWS</tt>, the results may be unpredictable. If the window frame is defined by <tt>RANGE</tt> or <tt>GROUPS</tt>, the results are same for each tuple.</p>
+</li>
+</ul>
+<p>This clause may have multiple <a href="#Ordering_term">ordering terms</a>. To reduce the number of ties, add additional <a href="#Ordering_term">ordering terms</a>.</p>
+<div class="section">
+<h5><a name="Note"></a>Note</h5>
+<p>This clause does not guarantee the overall order of the query results. To guarantee the order of the final results, use the query ORDER BY clause.</p></div></div>
+<div class="section">
+<h4><a name="Ordering_Term"></a><a name="Ordering_term" id="Ordering_term">Ordering Term</a></h4>
+
+<div>
+<div>
+<pre class="source">OrderingTerm ::= Expression ( &lt;ASC&gt; | &lt;DESC&gt; )?
+</pre></div></div>
+
+<p>The <b>ordering term</b> specifies an ordering expression and collation.</p>
+<p>This clause has the same syntax and semantics as the ordering term for queries. Refer to the <a href="#Order_By_clauses">ORDER BY Clauses</a> section for details.</p></div>
+<div class="section">
+<h4><a name="Window_Frame_Clause"></a><a name="Window_frame_clause" id="Window_frame_clause">Window Frame Clause</a></h4>
+
+<div>
+<div>
+<pre class="source">WindowFrameClause ::= ( &lt;ROWS&gt; | &lt;RANGE&gt; | &lt;GROUPS&gt; ) WindowFrameExtent
+</pre></div></div>
+
+<p>The <b>window frame clause</b> defines the window frame.</p>
+<p>This clause can be used with all <a href="builtins.html#AggregateFunctions">aggregate functions</a> and some <a href="builtins.html#WindowFunctions">window functions</a> &#x2014; refer to the descriptions of individual functions for more details.</p>
+<p>This clause is allowed only when the <a href="#Window_order_clause">window order clause</a> is present.</p>
+<p>This clause is optional.</p>
+<ul>
+
+<li>
+
+<p>If this clause is omitted and there is no <a href="#Window_order_clause">window order clause</a>, the window frame is the entire partition.</p>
+</li>
+<li>
+
+<p>If this clause is omitted but there is a <a href="#Window_order_clause">window order clause</a>, the window frame becomes all tuples in the partition preceding the current tuple and its peers &#x2014; the same as <tt>RANGE BETWEEN UNBOUNDED PRECEDING AND CURRENT ROW</tt>.</p>
+</li>
+</ul>
+<p>The window frame can be defined in the following ways:</p>
+<ul>
+
+<li>
+
+<p><tt>ROWS</tt>: Counts the exact number of tuples within the frame. If window ordering doesn&#x2019;t result in unique ordering, the function may produce unpredictable results. You can add a unique expression or more window ordering expressions to produce unique ordering.</p>
+</li>
+<li>
+
+<p><tt>RANGE</tt>: Looks for a value offset within the frame. The function produces deterministic results.</p>
+</li>
+<li>
+
+<p><tt>GROUPS</tt>: Counts all groups of tied rows within the frame. The function produces deterministic results.</p>
+</li>
+</ul>
+<div class="section">
+<h5><a name="Note"></a>Note</h5>
+<p>If this clause uses <tt>RANGE</tt> with either <tt>Expression PRECEDING</tt> or <tt>Expression FOLLOWING</tt>, the <a href="#Window_order_clause">window order clause</a> must have only a single ordering term.</p>
+<p>The ordering term expression must evaluate to a number.</p><!--
+The ordering term expression must evaluate to a number, a date, a time, or a
+datetime.
+If the ordering term expression evaluates to a date, a time, or a datetime, the
+expression in `Expression PRECEDING` or `Expression FOLLOWING` must evaluate to
+a duration.
+-->
+
+<p>If these conditions are not met, the window frame will be empty, which means the window function will return its default value: in most cases this is NULL, except for <tt>strict_count()</tt> or <tt>array_count()</tt>, whose default value is 0.</p>
+<p>This restriction does not apply when the window frame uses <tt>ROWS</tt> or <tt>GROUPS</tt>.</p></div>
+<div class="section">
+<h5><a name="Tip"></a>Tip</h5>
+<p>The <tt>RANGE</tt> window frame is commonly used to define window frames based on date or time.</p>
+<p>If you want to use <tt>RANGE</tt> with either <tt>Expression PRECEDING</tt> or <tt>Expression FOLLOWING</tt>, and you want to use an ordering expression based on date or time, the expression in <tt>Expression PRECEDING</tt> or <tt>Expression FOLLOWING</tt> must use a data type that can be added to the ordering expression.</p></div></div>
+<div class="section">
+<h4><a name="Window_Frame_Extent"></a><a name="Window_frame_extent" id="Window_frame_extent">Window Frame Extent</a></h4>
+
+<div>
+<div>
+<pre class="source">WindowFrameExtent ::= ( ( &lt;UNBOUNDED&gt; | Expression ) &lt;PRECEDING&gt; | &lt;CURRENT&gt; &lt;ROW&gt; ) |
+&lt;BETWEEN&gt;
+  ( &lt;UNBOUNDED&gt; &lt;PRECEDING&gt; | &lt;CURRENT&gt; &lt;ROW&gt; | Expression ( &lt;PRECEDING&gt; | &lt;FOLLOWING&gt; ) )
+&lt;AND&gt;
+  ( &lt;UNBOUNDED&gt; &lt;FOLLOWING&gt; | &lt;CURRENT&gt; &lt;ROW&gt; | Expression ( &lt;PRECEDING&gt; | &lt;FOLLOWING&gt; ) )
+</pre></div></div>
+
+<p>The <b>window frame extent clause</b> specifies the start point and end point of the window frame. The expression before <tt>AND</tt> is the start point and the expression after <tt>AND</tt> is the end point. If <tt>BETWEEN</tt> is omitted, you can only specify the start point; the end point becomes <tt>CURRENT ROW</tt>.</p>
+<p>The window frame end point can&#x2019;t be before the start point. If this clause violates this restriction explicitly, an error will result. If it violates this restriction implicitly, the window frame will be empty, which means the window function will return its default value: in most cases this is NULL, except for <tt>strict_count()</tt> or <tt>array_count()</tt>, whose default value is 0.</p>
+<p>Window frame extents that result in an explicit violation are:</p>
+<ul>
+
+<li>
+
+<p><tt>BETWEEN CURRENT ROW AND Expression PRECEDING</tt></p>
+</li>
+<li>
+
+<p><tt>BETWEEN Expression FOLLOWING AND Expression PRECEDING</tt></p>
+</li>
+<li>
+
+<p><tt>BETWEEN Expression FOLLOWING AND CURRENT ROW</tt></p>
+</li>
+</ul>
+<p>Window frame extents that result in an implicit violation are:</p>
+<ul>
+
+<li>
+
+<p><tt>BETWEEN UNBOUNDED PRECEDING AND Expression PRECEDING</tt> &#x2014; if <tt>Expression</tt> is too high, some tuples may generate an empty window frame.</p>
+</li>
+<li>
+
+<p><tt>BETWEEN Expression PRECEDING AND Expression PRECEDING</tt> &#x2014; if the second <tt>Expression</tt> is greater than or equal to the first <tt>Expression</tt>, all result sets will generate an empty window frame.</p>
+</li>
+<li>
+
+<p><tt>BETWEEN Expression FOLLOWING AND Expression FOLLOWING</tt> &#x2014; if the first <tt>Expression</tt> is greater than or equal to the second <tt>Expression</tt>, all result sets will generate an empty window frame.</p>
+</li>
+<li>
+
+<p><tt>BETWEEN Expression FOLLOWING AND UNBOUNDED FOLLOWING</tt> &#x2014; if <tt>Expression</tt> is too high, some tuples may generate an empty window frame.</p>
+</li>
+<li>
+
+<p>If the <a href="#Window_frame_exclusion">window frame exclusion clause</a> is present, any window frame specification may result in empty window frame.</p>
+</li>
+</ul>
+<p>The <tt>Expression</tt> must be a positive constant or an expression that evaluates as a positive number. For <tt>ROWS</tt> or <tt>GROUPS</tt>, the <tt>Expression</tt> must be an integer.</p></div>
+<div class="section">
+<h4><a name="Window_Frame_Exclusion"></a><a name="Window_frame_exclusion" id="Window_frame_exclusion">Window Frame Exclusion</a></h4>
+
+<div>
+<div>
+<pre class="source">WindowFrameExclusion ::= &lt;EXCLUDE&gt; ( &lt;CURRENT&gt; &lt;ROW&gt; | &lt;GROUP&gt; | &lt;TIES&gt; |
+&lt;NO&gt; &lt;OTHERS&gt; )
+</pre></div></div>
+
+<p>The <b>window frame exclusion clause</b> enables you to exclude specified tuples from the window frame.</p>
+<p>This clause can be used with all <a href="builtins.html#AggregateFunctions">aggregate functions</a> and some <a href="builtins.html#WindowFunctions">window functions</a> &#x2014; refer to the descriptions of individual functions for more details.</p>
+<p>This clause is allowed only when the <a href="#Window_frame_clause">window frame clause</a> is present.</p>
+<p>This clause is optional. If this clause is omitted, the default is no exclusion &#x2014; the same as <tt>EXCLUDE NO OTHERS</tt>.</p>
+<ul>
+
+<li>
+
+<p><tt>EXCLUDE CURRENT ROW</tt>: If the current tuple is still part of the window frame, it is removed from the window frame.</p>
+</li>
+<li>
+
+<p><tt>EXCLUDE GROUP</tt>: The current tuple and any peers of the current tuple are removed from the window frame.</p>
+</li>
+<li>
+
+<p><tt>EXCLUDE TIES</tt>: Any peers of the current tuple, but not the current tuple itself, are removed from the window frame.</p>
+</li>
+<li>
+
+<p><tt>EXCLUDE NO OTHERS</tt>: No additional tuples are removed from the window frame.</p>
+</li>
+</ul>
+<p>If the current tuple is already removed from the window frame, then it remains removed from the window frame.</p></div></div></div>
+<div class="section">
+<h2><a name="Subqueries" id="Subqueries">Subqueries</a></h2>
+<p>In the query language, an arbitrary subquery can appear anywhere that an expression can appear. Unlike SQL-92, as was just alluded to, the subqueries in a SELECT list or a boolean predicate need not return singleton, single-column relations. Instead, they may return arbitrary collections. For example, the following query is a variant of the prior group-by query examples; it retrieves an array of up to two &#x201c;dislike&#x201d; messages per user.</p>
+<div class="section">
+<div class="section">
+<div class="section">
+<h5><a name="Example"></a>Example</h5>
+
+<div>
+<div>
+<pre class="source">SELECT uid,
+       (SELECT VALUE m.msg
+        FROM msgs m
+        WHERE m.msg.message LIKE '%dislike%'
+        ORDER BY m.msg.messageId
+        LIMIT 2) AS msgs
+FROM GleambookMessages message
+GROUP BY message.authorId AS uid GROUP AS msgs(message AS msg);
+</pre></div></div>
+
+<p>For our sample data set, this query returns:</p>
+
+<div>
+<div>
+<pre class="source">[ {
+    &quot;msgs&quot;: [
+        {
+            &quot;senderLocation&quot;: [
+                41.66,
+                80.87
+            ],
+            &quot;inResponseTo&quot;: 4,
+            &quot;messageId&quot;: 2,
+            &quot;authorId&quot;: 1,
+            &quot;message&quot;: &quot; dislike x-phone its touch-screen is horrible&quot;
+        }
+    ],
+    &quot;uid&quot;: 1
+}, {
+    &quot;msgs&quot;: [
+
+    ],
+    &quot;uid&quot;: 2
+} ]
+</pre></div></div>
+
+<p>Note that a subquery, like a top-level <tt>SELECT</tt> statment, always returns a collection &#x2013; regardless of where within a query the subquery occurs &#x2013; and again, its result is never automatically cast into a scalar.</p></div></div></div></div>
+<div class="section">
+<h2><a name="Differences_from_SQL-92"></a><a name="Vs_SQL-92" id="Vs_SQL-92">Differences from SQL-92</a></h2>
+<p>The query language offers the following additional features beyond SQL-92:</p>
+<ul>
+
+<li>Fully composable and functional: A subquery can iterate over any intermediate collection and can appear anywhere in a query.</li>
+<li>Schema-free: The query language does not assume the existence of a static schema for any data that it processes.</li>
+<li>Correlated FROM terms: A right-side FROM term expression can refer to variables defined by FROM terms on its left.</li>
+<li>Powerful GROUP BY: In addition to a set of aggregate functions as in standard SQL, the groups created by the <tt>GROUP BY</tt> clause are directly usable in nested queries and/or to obtain nested results.</li>
+<li>Generalized SELECT clause: A SELECT clause can return any type of collection, while in SQL-92, a <tt>SELECT</tt> clause has to return a (homogeneous) collection of objects.</li>
+</ul>
+<p>The following matrix is a quick &#x201c;SQL-92 compatibility cheat sheet&#x201d; for the query language.</p>
+<table border="0" class="table table-striped">
+<thead>
+
+<tr class="a">
+<th> Feature </th>
+<th>  The query language </th>
+<th> SQL-92 </th>
+<th>  Why different?  </th></tr>
+</thead><tbody>
+
+<tr class="b">
+<td> SELECT * </td>
+<td> Returns nested objects </td>
+<td> Returns flattened concatenated objects </td>
+<td> Nested collections are 1st class citizens </td></tr>
+<tr class="a">
+<td> SELECT list </td>
+<td> order not preserved </td>
+<td> order preserved </td>
+<td> Fields in a JSON object are not ordered </td></tr>
+<tr class="b">
+<td> Subquery </td>
+<td> Returns a collection  </td>
+<td> The returned collection is cast into a scalar value if the subquery appears in a SELECT list or on one side of a comparison or as input to a function </td>
+<td> Nested collections are 1st class citizens </td></tr>
+<tr class="a">
+<td> LEFT OUTER JOIN </td>
+<td>  Fills in <tt>MISSING</tt>(s) for non-matches  </td>
+<td>   Fills in <tt>NULL</tt>(s) for non-matches    </td>
+<td> &#x201c;Absence&#x201d; is more appropriate than &#x201c;unknown&#x201d; here  </td></tr>
+<tr class="b">
+<td> UNION ALL       </td>
+<td> Allows heterogeneous inputs and output </td>
+<td> Input streams must be UNION-compatible and output field names are drawn from the first input stream </td>
+<td> Heterogenity and nested collections are common </td></tr>
+<tr class="a">
+<td> IN constant_expr </td>
+<td> The constant expression has to be an array or multiset, i.e., [..,..,&#x2026;] </td>
+<td> The constant collection can be represented as comma-separated items in a paren pair </td>
+<td> Nested collections are 1st class citizens </td></tr>
+<tr class="b">
+<td> String literal </td>
+<td> Double quotes or single quotes </td>
+<td> Single quotes only </td>
+<td> Double quoted strings are pervasive </td></tr>
+<tr class="a">
+<td> Delimited identifiers </td>
+<td> Backticks </td>
+<td> Double quotes </td>
+<td> Double quoted strings are pervasive </td></tr>
+</tbody>
+</table>
+<p>The following SQL-92 features are not implemented yet. However, the query language does not conflict with these features:</p>
+<ul>
+
+<li>CROSS JOIN, NATURAL JOIN, UNION JOIN</li>
+<li>RIGHT and FULL OUTER JOIN</li>
+<li>INTERSECT, EXCEPT, UNION with set semantics</li>
+<li>CAST expression</li>
+<li>COALESCE expression</li>
+<li>ALL and SOME predicates for linking to subqueries</li>
+<li>UNIQUE predicate (tests a collection for duplicates)</li>
+<li>MATCH predicate (tests for referential integrity)</li>
+<li>Row and Table constructors</li>
+<li>Preserved order for expressions in a SELECT list</li>
+</ul><!--
+ ! 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.
+ !-->
+
+<h1><a name="Errors" id="Errors">4. Errors</a></h1><!--
+ ! 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.
+ !-->
+
+<p>A query can potentially result in one of the following errors:</p>
+<ul>
+
+<li>syntax error,</li>
+<li>identifier resolution error,</li>
+<li>type error,</li>
+<li>resource error.</li>
+</ul>
+<p>If the query processor runs into any error, it will terminate the ongoing processing of the query and immediately return an error message to the client.</p></div>
+<div class="section">
+<h2><a name="Syntax_Errors"></a><a name="Syntax_errors" id="Syntax_errors">Syntax Errors</a></h2>
+<p>A valid query must satisfy the grammar rules of the query language. Otherwise, a syntax error will be raised.</p>
+<div class="section">
+<div class="section">
+<div class="section">
+<h5><a name="Example"></a>Example</h5>
+
+<div>
+<div>
+<pre class="source">SELECT *
+GleambookUsers user
+</pre></div></div>
+
+<p>Since the query misses a <tt>FROM</tt> keyword before the dataset <tt>GleambookUsers</tt>, we will get a syntax error as follows:</p>
+
+<div>
+<div>
+<pre class="source">Syntax error: In line 2 &gt;&gt;GleambookUsers user;&lt;&lt; Encountered &lt;IDENTIFIER&gt; \&quot;GleambookUsers\&quot; at column 1.
+</pre></div></div>
+</div>
+<div class="section">
+<h5><a name="Example"></a>Example</h5>
+
+<div>
+<div>
+<pre class="source">SELECT *
+FROM GleambookUsers user
+WHERE type=&quot;advertiser&quot;;
+</pre></div></div>
+
+<p>Since &#x201c;type&#x201d; is a reserved keyword in the query parser, we will get a syntax error as follows:</p>
+
+<div>
+<div>
+<pre class="source">Error: Syntax error: In line 3 &gt;&gt;WHERE type=&quot;advertiser&quot;;&lt;&lt; Encountered 'type' &quot;type&quot; at column 7.
+==&gt; WHERE type=&quot;advertiser&quot;;
+</pre></div></div>
+</div></div></div></div>
+<div class="section">
+<h2><a name="Identifier_Resolution_Errors"></a><a name="Identifier_resolution_errors" id="Identifier_resolution_errors">Identifier Resolution Errors</a></h2>
+<p>Referring to an undefined identifier can cause an error if the identifier cannot be successfully resolved as a valid field access.</p>
+<div class="section">
+<div class="section">
+<div class="section">
+<h5><a name="Example"></a>Example</h5>
+
+<div>
+<div>
+<pre class="source">SELECT *
+FROM GleambookUser user;
+</pre></div></div>
+
+<p>If we have a typo as above in &#x201c;GleambookUsers&#x201d; that misses the dataset name&#x2019;s ending &#x201c;s&#x201d;, we will get an identifier resolution error as follows:</p>
+
+<div>
+<div>
+<pre class="source">Error: Cannot find dataset GleambookUser in dataverse Default nor an alias with name GleambookUser!
+</pre></div></div>
+</div>
+<div class="section">
+<h5><a name="Example"></a>Example</h5>
+
+<div>
+<div>
+<pre class="source">SELECT name, message
+FROM GleambookUsers u JOIN GleambookMessages m ON m.authorId = u.id;
+</pre></div></div>
+
+<p>If the compiler cannot figure out how to resolve an unqualified field name, which will occur if there is more than one variable in scope (e.g., <tt>GleambookUsers u</tt> and <tt>GleambookMessages m</tt> as above), we will get an identifier resolution error as follows:</p>
+
+<div>
+<div>
+<pre class="source">Error: Cannot resolve ambiguous alias reference for undefined identifier name
+</pre></div></div>
+</div></div></div></div>
+<div class="section">
+<h2><a name="Type_Errors"></a><a name="Type_errors" id="Type_errors">Type Errors</a></h2>
+<p>The query compiler does type checks based on its available type information. In addition, the query runtime also reports type errors if a data model instance it processes does not satisfy the type requirement.</p>
+<div class="section">
+<div class="section">
+<div class="section">
+<h5><a name="Example"></a>Example</h5>
+
+<div>
+<div>
+<pre class="source">abs(&quot;123&quot;);
+</pre></div></div>
+
+<p>Since function <tt>abs</tt> can only process numeric input values, we will get a type error as follows:</p>
+
+<div>
+<div>
+<pre class="source">Error: Type mismatch: function abs expects its 1st input parameter to be of type tinyint, smallint, integer, bigint, float or double, but the actual input type is string
+</pre></div></div>
+</div></div></div></div>
+<div class="section">
+<h2><a name="Resource_Errors"></a><a name="Resource_errors" id="Resource_errors">Resource Errors</a></h2>
+<p>A query can potentially exhaust system resources, such as the number of open files and disk spaces. For instance, the following two resource errors could be potentially be seen when running the system:</p>
+
+<div>
+<div>
+<pre class="source">Error: no space left on device
+Error: too many open files
+</pre></div></div>
+
+<p>The &#x201c;no space left on device&#x201d; issue usually can be fixed by cleaning up disk spaces and reserving more disk spaces for the system. The &#x201c;too many open files&#x201d; issue usually can be fixed by a system administrator, following the instructions <a class="externalLink" href="https://easyengine.io/tutorials/linux/increase-open-files-limit/">here</a>.</p><!--
+ ! 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.
+ !-->
+
+<h1><a name="DDL_and_DML_statements" id="DDL_and_DML_statements">5. DDL and DML statements</a></h1>
+
+<div>
+<div>
+<pre class="source">Statement ::= ( ( SingleStatement )? ( &quot;;&quot; )+ )* &lt;EOF&gt;
+SingleStatement ::= DatabaseDeclaration
+                  | FunctionDeclaration
+                  | CreateStatement
+                  | DropStatement
+                  | LoadStatement
+                  | SetStatement
+                  | InsertStatement
+                  | DeleteStatement
+                  | Query
+</pre></div></div>
+
+<p>In addition to queries, an implementation of the query language needs to support statements for data definition and manipulation purposes as well as controlling the context to be used in evaluating query expressions. This section details the DDL and DML statements supported in the query language as realized today in Apache AsterixDB.</p><!--
+ ! 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.
+ !-->
+</div>
+<div class="section">
+<h2><a name="Lifecycle_Management_Statements"></a><a name="Lifecycle_management_statements" id="Lifecycle_management_statements">Lifecycle Management Statements</a></h2>
+
+<div>
+<div>
+<pre class="source">CreateStatement ::= &quot;CREATE&quot; ( DatabaseSpecification
+                             | TypeSpecification
+                             | DatasetSpecification
+                             | IndexSpecification
+                             | SynonymSpecification
+                             | FunctionSpecification )
+
+QualifiedName       ::= Identifier ( &quot;.&quot; Identifier )?
+DoubleQualifiedName ::= Identifier &quot;.&quot; Identifier ( &quot;.&quot; Identifier )?
+</pre></div></div>
+
+<p>The CREATE statement is used for creating dataverses as well as other persistent artifacts in a dataverse. It can be used to create new dataverses, datatypes, datasets, indexes, and user-defined query functions.</p>
+<div class="section">
+<h3><a name="Dataverses" id="Dataverses"> Dataverses</a></h3>
+
+<div>
+<div>
+<pre class="source">DatabaseSpecification ::= &quot;DATAVERSE&quot; Identifier IfNotExists
+</pre></div></div>
+
+<p>The CREATE DATAVERSE statement is used to create new dataverses. To ease the authoring of reusable query scripts, an optional IF NOT EXISTS clause is included to allow creation to be requested either unconditionally or only if the dataverse does not already exist. If this clause is absent, an error is returned if a dataverse with the indicated name already exists.</p>
+<p>The following example creates a new dataverse named TinySocial if one does not already exist.</p>
+<div class="section">
+<div class="section">
+<h5><a name="Example"></a>Example</h5>
+
+<div>
+<div>
+<pre class="source">CREATE DATAVERSE TinySocial IF NOT EXISTS;
+</pre></div></div>
+</div></div></div>
+<div class="section">
+<h3><a name="Types" id="Types"> Types</a></h3>
+
+<div>
+<div>
+<pre class="source">TypeSpecification    ::= &quot;TYPE&quot; FunctionOrTypeName IfNotExists &quot;AS&quot; ObjectTypeDef
+FunctionOrTypeName   ::= QualifiedName
+IfNotExists          ::= ( &lt;IF&gt; &lt;NOT&gt; &lt;EXISTS&gt; )?
+TypeExpr             ::= ObjectTypeDef | TypeReference | ArrayTypeDef | MultisetTypeDef
+ObjectTypeDef        ::= ( &lt;CLOSED&gt; | &lt;OPEN&gt; )? &quot;{&quot; ( ObjectField ( &quot;,&quot; ObjectField )* )? &quot;}&quot;
+ObjectField          ::= Identifier &quot;:&quot; ( TypeExpr ) ( &quot;?&quot; )?
+NestedField          ::= Identifier ( &quot;.&quot; Identifier )*
+IndexField           ::= NestedField ( &quot;:&quot; TypeReference )?
+TypeReference        ::= Identifier
+ArrayTypeDef         ::= &quot;[&quot; ( TypeExpr ) &quot;]&quot;
+MultisetTypeDef      ::= &quot;{{&quot; ( TypeExpr ) &quot;}}&quot;
+</pre></div></div>
+
+<p>The CREATE TYPE statement is used to create a new named datatype. This type can then be used to create stored collections or utilized when defining one or more other datatypes. Much more information about the data model is available in the <a href="../datamodel.html">data model reference guide</a>. A new type can be a object type, a renaming of another type, an array type, or a multiset type. A object type can be defined as being either open or closed. Instances of a closed object type are not permitted to contain fields other than those specified in the create type statement. Instances of an open object type may carry additional fields, and open is the default for new types if neither option is specified.</p>
+<p>The following example creates a new object type called GleambookUser type. Since it is defined as (defaulting to) being an open type, instances will be permitted to contain more than what is specified in the type definition. The first four fields are essentially traditional typed name/value pairs (much like SQL fields). The friendIds field is a multiset of integers. The employment field is an array of instances of another named object type, EmploymentType.</p>
+<div class="section">
+<div class="section">
+<h5><a name="Example"></a>Example</h5>
+
+<div>
+<div>
+<pre class="source">CREATE TYPE GleambookUserType AS {
+  id:         int,
+  alias:      string,
+  name:       string,
+  userSince: datetime,
+  friendIds: {{ int }},
+  employment: [ EmploymentType ]
+};
+</pre></div></div>
+
+<p>The next example creates a new object type, closed this time, called MyUserTupleType. Instances of this closed type will not be permitted to have extra fields, although the alias field is marked as optional and may thus be NULL or MISSING in legal instances of the type. Note that the type of the id field in the example is UUID. This field type can be used if you want to have this field be an autogenerated-PK field. (Refer to the Datasets section later for more details on such fields.)</p></div>
+<div class="section">
+<h5><a name="Example"></a>Example</h5>
+
+<div>
+<div>
+<pre class="source">CREATE TYPE MyUserTupleType AS CLOSED {
+  id:         uuid,
+  alias:      string?,
+  name:       string
+};
+</pre></div></div>
+</div></div></div>
+<div class="section">
+<h3><a name="Datasets" id="Datasets"> Datasets</a></h3>
+
+<div>
+<div>
+<pre class="source">DatasetSpecification ::= ( &lt;INTERNAL&gt; )? &lt;DATASET&gt; QualifiedName &quot;(&quot; QualifiedName &quot;)&quot; IfNotExists
+                           PrimaryKey ( &lt;ON&gt; Identifier )? ( &lt;HINTS&gt; Properties )?
+                           ( &quot;USING&quot; &quot;COMPACTION&quot; &quot;POLICY&quot; CompactionPolicy ( Configuration )? )?
+                           ( &lt;WITH&gt; &lt;FILTER&gt; &lt;ON&gt; Identifier )?
+                          |
+                           &lt;EXTERNAL&gt; &lt;DATASET&gt; QualifiedName &quot;(&quot; QualifiedName &quot;)&quot; IfNotExists &lt;USING&gt; AdapterName
+                           Configuration ( &lt;HINTS&gt; Properties )?
+                           ( &lt;USING&gt; &lt;COMPACTION&gt; &lt;POLICY&gt; CompactionPolicy ( Configuration )? )?
+AdapterName          ::= Identifier
+Configuration        ::= &quot;(&quot; ( KeyValuePair ( &quot;,&quot; KeyValuePair )* )? &quot;)&quot;
+KeyValuePair         ::= &quot;(&quot; StringLiteral &quot;=&quot; StringLiteral &quot;)&quot;
+Properties           ::= ( &quot;(&quot; Property ( &quot;,&quot; Property )* &quot;)&quot; )?
+Property             ::= Identifier &quot;=&quot; ( StringLiteral | IntegerLiteral )
+FunctionSignature    ::= FunctionOrTypeName &quot;@&quot; IntegerLiteral
+PrimaryKey           ::= &lt;PRIMARY&gt; &lt;KEY&gt; NestedField ( &quot;,&quot; NestedField )* ( &lt;AUTOGENERATED&gt; )?
+CompactionPolicy     ::= Identifier
+</pre></div></div>
+
+<p>The CREATE DATASET statement is used to create a new dataset. Datasets are named, multisets of object type instances; they are where data lives persistently and are the usual targets for queries. Datasets are typed, and the system ensures that their contents conform to their type definitions. An Internal dataset (the default kind) is a dataset whose content lives within and is managed by the system. It is required to have a specified unique primary key field which uniquely identifies the contained objects. (The primary key is also used in secondary indexes to identify the indexed primary data objects.)</p>
+<p>Internal datasets contain several advanced options that can be specified when appropriate. One such option is that random primary key (UUID) values can be auto-generated by declaring the field to be UUID and putting &#x201c;AUTOGENERATED&#x201d; after the &#x201c;PRIMARY KEY&#x201d; identifier. In this case, unlike other non-optional fields, a value for the auto-generated PK field should not be provided at insertion time by the user since each object&#x2019;s primary key field value will be auto-generated by the system.</p>
+<p>Another advanced option, when creating an Internal dataset, is to specify the merge policy to control which of the underlying LSM storage components to be merged. (The system supports Log-Structured Merge tree based physical storage for Internal datasets.) Currently the system supports four different component merging policies that can be chosen per dataset: no-merge, constant, prefix, and correlated-prefix. The no-merge policy simply never merges disk components. The constant policy merges disk components when the number of components reaches a constant number k that can be configured by the user. The prefix policy relies on both component sizes and the number of components to decide which components to merge. It works by first trying to identify the smallest ordered (oldest to newest) sequence of components such that the sequence does not contain a single component that exceeds some threshold size M and that either the sum of the component&#x2019;s sizes exceeds M or the number of components in the sequence exceeds another threshold C. If such a sequence exists, the components in the sequence are merged together to form a single component. Finally, the correlated-prefix policy is similar to the prefix policy, but it delegates the decision of merging the disk components of all the indexes in a dataset to the primary index. When the correlated-prefix policy decides that the primary index needs to be merged (using the same decision criteria as for the prefix policy), then it will issue successive merge requests on behalf of all other indexes associated with the same dataset. The system&#x2019;s default policy is the prefix policy except when there is a filter on a dataset, where the preferred policy for filters is the correlated-prefix.</p>
+<p>Another advanced option shown in the syntax above, related to performance and mentioned above, is that a <b>filter</b> can optionally be created on a field to further optimize range queries with predicates on the filter&#x2019;s field. Filters allow some range queries to avoid searching all LSM components when the query conditions match the filter. (Refer to <a href="../filters.html">Filter-Based LSM Index Acceleration</a> for more information about filters.)</p>
+<p>An External dataset, in contrast to an Internal dataset, has data stored outside of the system&#x2019;s control. Files living in HDFS or in the local filesystem(s) of a cluster&#x2019;s nodes are currently supported. External dataset support allows queries to treat foreign data as though it were stored in the system, making it possible to query &#x201c;legacy&#x201d; file data (for example, Hive data) without having to physically import it. When defining an External dataset, an appropriate adapter type must be selected for the desired external data. (See the <a href="../externaldata.html">Guide to External Data</a> for more information on the available adapters.)</p>
+<p>The following example creates an Internal dataset for storing FacefookUserType objects. It specifies that their id field is their primary key.</p>
+<div class="section">
+<h4><a name="Example"></a>Example</h4>
+
+<div>
+<div>
+<pre class="source">CREATE INTERNAL DATASET GleambookUsers(GleambookUserType) PRIMARY KEY id;
+</pre></div></div>
+
+<p>The next example creates another Internal dataset (the default kind when no dataset kind is specified) for storing MyUserTupleType objects. It specifies that the id field should be used as the primary key for the dataset. It also specifies that the id field is an auto-generated field, meaning that a randomly generated UUID value should be assigned to each incoming object by the system. (A user should therefore not attempt to provide a value for this field.) Note that the id field&#x2019;s declared type must be UUID in this case.</p></div>
+<div class="section">
+<h4><a name="Example"></a>Example</h4>
+
+<div>
+<div>
+<pre class="source">CREATE DATASET MyUsers(MyUserTupleType) PRIMARY KEY id AUTOGENERATED;
+</pre></div></div>
+
+<p>The next example creates an External dataset for querying LineItemType objects. The choice of the <tt>hdfs</tt> adapter means that this dataset&#x2019;s data actually resides in HDFS. The example CREATE statement also provides parameters used by the hdfs adapter: the URL and path needed to locate the data in HDFS and a description of the data format.</p></div>
+<div class="section">
+<h4><a name="Example"></a>Example</h4>
+
+<div>
+<div>
+<pre class="source">CREATE EXTERNAL DATASET LineItem(LineItemType) USING hdfs (
+  (&quot;hdfs&quot;=&quot;hdfs://HOST:PORT&quot;),
+  (&quot;path&quot;=&quot;HDFS_PATH&quot;),
+  (&quot;input-format&quot;=&quot;text-input-format&quot;),
+  (&quot;format&quot;=&quot;delimited-text&quot;),
+  (&quot;delimiter&quot;=&quot;|&quot;));
+</pre></div></div>
+</div></div>
+<div class="section">
+<h3><a name="Indices" id="Indices">Indices</a></h3>
+
+<div>
+<div>
+<pre class="source">IndexSpecification ::= (&lt;INDEX&gt; Identifier IfNotExists &lt;ON&gt; QualifiedName
+                       &quot;(&quot; ( IndexField ) ( &quot;,&quot; IndexField )* &quot;)&quot; (&lt;TYPE&gt; IndexType)? (&lt;ENFORCED&gt;)?)
+                       |
+                       &lt;PRIMARY&gt; &lt;INDEX&gt; Identifier? IfNotExists &lt;ON&gt; QualifiedName (&lt;TYPE&gt; &lt;BTREE&gt;)?
+IndexType          ::= &lt;BTREE&gt; | &lt;RTREE&gt; | &lt;KEYWORD&gt; | &lt;NGRAM&gt; &quot;(&quot; IntegerLiteral &quot;)&quot;
+</pre></div></div>
+
+<p>The CREATE INDEX statement creates a secondary index on one or more fields of a specified dataset. Supported index types include <tt>BTREE</tt> for totally ordered datatypes, <tt>RTREE</tt> for spatial data, and <tt>KEYWORD</tt> and <tt>NGRAM</tt> for textual (string) data. An index can be created on a nested field (or fields) by providing a valid path expression as an index field identifier.</p>
+<p>An indexed field is not required to be part of the datatype associated with a dataset if the dataset&#x2019;s datatype is declared as open <b>and</b> if the field&#x2019;s type is provided along with its name and if the <tt>ENFORCED</tt> keyword is specified at the end of the index definition. <tt>ENFORCING</tt> an open field introduces a check that makes sure that the actual type of the indexed field (if the optional field exists in the object) always matches this specified (open) field type.</p>
+<p>The following example creates a btree index called gbAuthorIdx on the authorId field of the GleambookMessages dataset. This index can be useful for accelerating exact-match queries, range search queries, and joins involving the author-id field.</p>
+<div class="section">
+<h4><a name="Example"></a>Example</h4>
+
+<div>
+<div>
+<pre class="source">CREATE INDEX gbAuthorIdx ON GleambookMessages(authorId) TYPE BTREE;
+</pre></div></div>
+
+<p>The following example creates an open btree index called gbSendTimeIdx on the (non-declared) <tt>sendTime</tt> field of the GleambookMessages dataset having datetime type. This index can be useful for accelerating exact-match queries, range search queries, and joins involving the <tt>sendTime</tt> field. The index is enforced so that records that do not have the <tt>sendTime</tt> field or have a mismatched type on the field cannot be inserted into the dataset.</p></div>
+<div class="section">
+<h4><a name="Example"></a>Example</h4>
+
+<div>
+<div>
+<pre class="source">CREATE INDEX gbSendTimeIdx ON GleambookMessages(sendTime: datetime?) TYPE BTREE ENFORCED;
+</pre></div></div>
+
+<p>The following example creates an open btree index called gbReadTimeIdx on the (non-declared) <tt>readTime</tt> field of the GleambookMessages dataset having datetime type. This index can be useful for accelerating exact-match queries, range search queries, and joins involving the <tt>readTime</tt> field. The index is not enforced so that records that do not have the <tt>readTime</tt> field or have a mismatched type on the field can still be inserted into the dataset.</p></div>
+<div class="section">
+<h4><a name="Example"></a>Example</h4>
+
+<div>
+<div>
+<pre class="source">CREATE INDEX gbReadTimeIdx ON GleambookMessages(readTime: datetime?);
+</pre></div></div>
+
+<p>The following example creates a btree index called crpUserScrNameIdx on screenName, a nested field residing within a object-valued user field in the ChirpMessages dataset. This index can be useful for accelerating exact-match queries, range search queries, and joins involving the nested screenName field. Such nested fields must be singular, i.e., one cannot index through (or on) an array-valued field.</p></div>
+<div class="section">
+<h4><a name="Example"></a>Example</h4>
+
+<div>
+<div>
+<pre class="source">CREATE INDEX crpUserScrNameIdx ON ChirpMessages(user.screenName) TYPE BTREE;
+</pre></div></div>
+
+<p>The following example creates an rtree index called gbSenderLocIdx on the sender-location field of the GleambookMessages dataset. This index can be useful for accelerating queries that use the <a href="functions.html#spatial-intersect"><tt>spatial-intersect</tt> function</a> in a predicate involving the sender-location field.</p></div>
+<div class="section">
+<h4><a name="Example"></a>Example</h4>
+
+<div>
+<div>
+<pre class="source">CREATE INDEX gbSenderLocIndex ON GleambookMessages(&quot;sender-location&quot;) TYPE RTREE;
+</pre></div></div>
+
+<p>The following example creates a 3-gram index called fbUserIdx on the name field of the GleambookUsers dataset. This index can be used to accelerate some similarity or substring maching queries on the name field. For details refer to the document on <a href="similarity.html#NGram_Index">similarity queries</a>.</p></div>
+<div class="section">
+<h4><a name="Example"></a>Example</h4>
+
+<div>
+<div>
+<pre class="source">CREATE INDEX fbUserIdx ON GleambookUsers(name) TYPE NGRAM(3);
+</pre></div></div>
+
+<p>The following example creates a keyword index called fbMessageIdx on the message field of the GleambookMessages dataset. This keyword index can be used to optimize queries with token-based similarity predicates on the message field. For details refer to the document on <a href="similarity.html#Keyword_Index">similarity queries</a>.</p></div>
+<div class="section">
+<h4><a name="Example"></a>Example</h4>
+
+<div>
+<div>
+<pre class="source">CREATE INDEX fbMessageIdx ON GleambookMessages(message) TYPE KEYWORD;
+</pre></div></div>
+
+<p>The following example creates a special secondary index which holds only the primary keys. This index is useful for speeding up aggregation queries which involve only primary keys. The name of the index is optional. If the name is not specified, the system will generate one. When the user would like to drop this index, the metadata can be queried to find the system-generated name.</p></div>
+<div class="section">
+<h4><a name="Example"></a>Example</h4>
+
+<div>
+<div>
+<pre class="source">CREATE PRIMARY INDEX gb_pk_idx ON GleambookMessages;
+</pre></div></div>
+
+<p>An example query that can be accelerated using the primary-key index:</p>
+
+<div>
+<div>
+<pre class="source">SELECT COUNT(*) FROM GleambookMessages;
+</pre></div></div>
+
+<p>To look up the the above primary-key index, issue the following query:</p>
+
+<div>
+<div>
+<pre class="source">SELECT VALUE i
+FROM Metadata.`Index` i
+WHERE i.DataverseName = &quot;TinySocial&quot; AND i.DatasetName = &quot;GleambookMessages&quot;;
+</pre></div></div>
+
+<p>The query returns:</p>
+
+<div>
+<div>
+<pre class="source">[ { &quot;DataverseName&quot;: &quot;TinySocial&quot;, &quot;DatasetName&quot;: &quot;GleambookMessages&quot;, &quot;IndexName&quot;: &quot;GleambookMessages&quot;, &quot;IndexStructure&quot;: &quot;BTREE&quot;, &quot;SearchKey&quot;: [ [ &quot;messageId&quot; ] ], &quot;IsPrimary&quot;: true, &quot;Timestamp&quot;: &quot;Wed Nov 07 17:25:11 PST 2018&quot;, &quot;PendingOp&quot;: 0 }
+, { &quot;DataverseName&quot;: &quot;TinySocial&quot;, &quot;DatasetName&quot;: &quot;GleambookMessages&quot;, &quot;IndexName&quot;: &quot;gb_pk_idx&quot;, &quot;IndexStructure&quot;: &quot;BTREE&quot;, &quot;SearchKey&quot;: [  ], &quot;IsPrimary&quot;: false, &quot;Timestamp&quot;: &quot;Wed Nov 07 17:25:11 PST 2018&quot;, &quot;PendingOp&quot;: 0 }
+ ]
+</pre></div></div>
+
+<p>Remember that <tt>CREATE PRIMARY INDEX</tt> creates a secondary index. That is the reason the <tt>IsPrimary</tt> field is false. The primary-key index can be identified by the fact that the <tt>SearchKey</tt> field is empty since it only contains primary key fields.<!--
+! 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.
+!--></p></div></div>
+<div class="section">
+<h3><a name="Functions" id="Functions"> Functions</a></h3>
+<p>The CREATE FUNCTION statement creates a <b>named</b> function that can then be used and reused in queries. The body of a function can be any query expression involving the function&#x2019;s parameters.</p>
+
+<div>
+<div>
+<pre class="source">FunctionSpecification ::= &quot;FUNCTION&quot; FunctionOrTypeName IfNotExists ParameterList &quot;{&quot; Expression &quot;}&quot;
+</pre></div></div>
+
+<p>The following is an example of a CREATE FUNCTION statement which is similar to our earlier DECLARE FUNCTION example. It differs from that example in that it results in a function that is persistently registered by name in the specified dataverse (the current dataverse being used, if not otherwise specified).</p>
+<div class="section">
+<div class="section">
+<h5><a name="Example"></a>Example</h5>
+
+<div>
+<div>
+<pre class="source">CREATE FUNCTION friendInfo(userId) {
+    (SELECT u.id, u.name, len(u.friendIds) AS friendCount
+     FROM GleambookUsers u
+     WHERE u.id = userId)[0]
+ };
+</pre></div></div>
+</div></div></div>
+<div class="section">
+<h3><a name="Synonyms" id="Synonyms"> Synonyms</a></h3>
+
+<div>
+<div>
+<pre class="source">SynonymSpecification ::= &quot;SYNONYM&quot; QualifiedName &quot;FOR&quot; QualifiedName IfNotExists
+</pre></div></div>
+
+<p>The CREATE SYNONYM statement creates a synonym for a given dataset. This synonym may be used used instead of the dataset name in SELECT, INSERT, UPSERT, DELETE, and LOAD statements. The target dataset does not need to exist when the synonym is created.</p>
+<div class="section">
+<div class="section">
+<h5><a name="Example"></a>Example</h5>
+
+<div>
+<div>
+<pre class="source">CREATE DATASET GleambookUsers(GleambookUserType) PRIMARY KEY id;
+
+CREATE SYNONYM GleambookUsersSynonym FOR GleambookUsers;
+
+SELECT * FROM GleambookUsersSynonym;
+</pre></div></div>
+
+<p>More information on how synonyms are resolved can be found in the appendix section on Variable Resolution.</p></div></div></div>
+<div class="section">
+<h3><a name="Removal" id="Removal"> Removal</a></h3>
+
+<div>
+<div>
+<pre class="source">DropStatement       ::= &quot;DROP&quot; ( &quot;DATAVERSE&quot; Identifier IfExists
+                               | &quot;TYPE&quot; FunctionOrTypeName IfExists
+                               | &quot;DATASET&quot; QualifiedName IfExists
+                               | &quot;INDEX&quot; DoubleQualifiedName IfExists
+                               | &quot;SYNONYM&quot; QualifiedName IfExists
+                               | &quot;FUNCTION&quot; FunctionSignature IfExists )
+IfExists            ::= ( &quot;IF&quot; &quot;EXISTS&quot; )?
+</pre></div></div>
+
+<p>The DROP statement is the inverse of the CREATE statement. It can be used to drop dataverses, datatypes, datasets, indexes, functions, and synonyms.</p>
+<p>The following examples illustrate some uses of the DROP statement.</p>
+<div class="section">
+<div class="section">
+<h5><a name="Example"></a>Example</h5>
+
+<div>
+<div>
+<pre class="source">DROP DATASET GleambookUsers IF EXISTS;
+
+DROP INDEX GleambookMessages.gbSenderLocIndex;
+
+DROP TYPE TinySocial2.GleambookUserType;
+
+DROP FUNCTION friendInfo@1;
+
+DROP SYNONYM GleambookUsersSynonym;
+
+DROP DATAVERSE TinySocial;
+</pre></div></div>
+
+<p>When an artifact is dropped, it will be droppped from the current dataverse if none is specified (see the DROP DATASET example above) or from the specified dataverse (see the DROP TYPE example above) if one is specified by fully qualifying the artifact name in the DROP statement. When specifying an index to drop, the index name must be qualified by the dataset that it indexes. When specifying a function to drop, since the query language allows functions to be overloaded by their number of arguments, the identifying name of the function to be dropped must explicitly include that information. (<tt>friendInfo@1</tt> above denotes the 1-argument function named friendInfo in the current dataverse.)</p></div></div></div>
+<div class="section">
+<h3><a name="Load_Statement"></a><a name="Load_statement" id="Load_statement">Load Statement</a></h3>
+
+<div>
+<div>
+<pre class="source">LoadStatement  ::= &lt;LOAD&gt; &lt;DATASET&gt; QualifiedName &lt;USING&gt; AdapterName Configuration ( &lt;PRE-SORTED&gt; )?
+</pre></div></div>
+
+<p>The LOAD statement is used to initially populate a dataset via bulk loading of data from an external file. An appropriate adapter must be selected to handle the nature of the desired external data. The LOAD statement accepts the same adapters and the same parameters as discussed earlier for External datasets. (See the <a href="externaldata.html">guide to external data</a> for more information on the available adapters.) If a dataset has an auto-generated primary key field, the file to be imported should not include that field in it.</p>
+<p>The target dataset name may be a synonym introduced by CREATE SYNONYM statement.</p>
+<p>The following example shows how to bulk load the GleambookUsers dataset from an external file containing data that has been prepared in ADM (Asterix Data Model) format.</p>
+<div class="section">
+<div class="section">
+<h5><a name="Example"></a>Example</h5>
+
+<div>
+<div>
+<pre class="source"> LOAD DATASET GleambookUsers USING localfs
+    ((&quot;path&quot;=&quot;127.0.0.1:///Users/bignosqlfan/tinysocialnew/gbu.adm&quot;),(&quot;format&quot;=&quot;adm&quot;));
+</pre></div></div>
+<!--
+ ! 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.
+ !-->
+</div></div></div></div>
+<div class="section">
+<h2><a name="Modification_statements" id="Modification_statements">Modification statements</a></h2>
+<div class="section">
+<h3><a name="INSERTs"></a><a name="Inserts" id="Inserts">INSERTs</a></h3>
+
+<div>
+<div>
+<pre class="source">InsertStatement ::= &lt;INSERT&gt; &lt;INTO&gt; QualifiedName Query
+</pre></div></div>
+
+<p>The INSERT statement is used to insert new data into a dataset. The data to be inserted comes from a query expression. This expression can be as simple as a constant expression, or in general it can be any legal query. In case the dataset has an auto-generated primary key, when performing an INSERT operation, the system allows the user to manually add the auto-generated key field in the INSERT statement, or skip that field and the system will automatically generate it and add it. However, it is important to note that if the a record already exists in the dataset with the auto-generated key provided by the user, then that operation is going to fail. As a general rule, insertion will fail if the dataset already has data with the primary key value(s) being inserted.</p>
+<p>Inserts are processed transactionally by the system. The transactional scope of each insert transaction is the insertion of a single object plus its affiliated secondary index entries (if any). If the query part of an insert returns a single object, then the INSERT statement will be a single, atomic transaction. If the query part returns multiple objects, each object being inserted will be treated as a separate tranaction.</p>
+<p>The target dataset name may be a synonym introduced by CREATE SYNONYM statement.</p>
+<p>The following example illustrates a query-based insertion.</p>
+<div class="section">
+<div class="section">
+<h5><a name="Example"></a>Example</h5>
+
+<div>
+<div>
+<pre class="source">INSERT INTO UsersCopy (SELECT VALUE user FROM GleambookUsers user)
+</pre></div></div>
+</div></div></div>
+<div class="section">
+<h3><a name="UPSERTs"></a><a name="Upserts" id="Upserts">UPSERTs</a></h3>
+
+<div>
+<div>
+<pre class="source">UpsertStatement ::= &lt;UPSERT&gt; &lt;INTO&gt; QualifiedName Query
+</pre></div></div>
+
+<p>The UPSERT statement syntactically mirrors the INSERT statement discussed above. The difference lies in its semantics, which for UPSERT are &#x201c;add or replace&#x201d; instead of the INSERT &#x201c;add if not present, else error&#x201d; semantics. Whereas an INSERT can fail if another object already exists with the specified key, the analogous UPSERT will replace the previous object&#x2019;s value with that of the new object in such cases. Like the INSERT statement, the system allows the user to manually provide the auto-generated key for datasets with an auto-generated key as its primary key. This operation will insert the record if no record with that key already exists, but if a record with the key already exists, then the operation will be converted to a replace/update operation.</p>
+<p>The target dataset name may be a synonym introduced by CREATE SYNONYM statement.</p>
+<p>The following example illustrates a query-based upsert operation.</p>
+<div class="section">
+<div class="section">
+<h5><a name="Example"></a>Example</h5>
+
+<div>
+<div>
+<pre class="source">UPSERT INTO UsersCopy (SELECT VALUE user FROM GleambookUsers user)
+</pre></div></div>
+
+<p>*Editor&#x2019;s note: Upserts currently work in AQL but are not yet enabled (at the moment) in the current query language.</p></div></div></div>
+<div class="section">
+<h3><a name="DELETEs"></a><a name="Deletes" id="Deletes">DELETEs</a></h3>
+
+<div>
+<div>
+<pre class="source">DeleteStatement ::= &lt;DELETE&gt; &lt;FROM&gt; QualifiedName ( ( &lt;AS&gt; )? Variable )? ( &lt;WHERE&gt; Expression )?
+</pre></div></div>
+
+<p>The DELETE statement is used to delete data from a target dataset. The data to be deleted is identified by a boolean expression involving the variable bound to the target dataset in the DELETE statement.</p>
+<p>Deletes are processed transactionally by the system. The transactional scope of each delete transaction is the deletion of a single object plus its affiliated secondary index entries (if any). If the boolean expression for a delete identifies a single object, then the DELETE statement itself will be a single, atomic transaction. If the expression identifies multiple objects, then each object deleted will be handled as a separate transaction.</p>
+<p>The target dataset name may be a synonym introduced by CREATE SYNONYM statement.</p>
+<p>The following examples illustrate single-object deletions.</p>
+<div class="section">
+<div class="section">
+<h5><a name="Example"></a>Example</h5>
+
+<div>
+<div>
+<pre class="source">DELETE FROM GleambookUsers user WHERE user.id = 8;
+</pre></div></div>
+</div>
+<div class="section">
+<h5><a name="Example"></a>Example</h5>
+
+<div>
+<div>
+<pre class="source">DELETE FROM GleambookUsers WHERE id = 5;
+</pre></div></div>
+<!--
+ ! 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.
+ !-->
+
+<h1><a name="Reserved_keywords" id="Reserved_keywords">Appendix 1. Reserved keywords</a></h1><!--
+ ! 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.
+ !-->
+
+<p>All reserved keywords are listed in the following table:</p>
+<table border="0" class="table table-striped">
+<thead>
+
+<tr class="a">
+<th>     </th>
+<th>     </th>
+<th>       </th>
+<th>    </th>
+<th>     </th>
+<th>    </th></tr>
+</thead><tbody>
+
+<tr class="b">
+<td> AND </td>
+<td> ANY </td>
+<td> APPLY </td>
+<td> AS </td>
+<td> ASC </td>
+<td> AT </td></tr>
+<tr class="a">
+<td> AUTOGENERATED </td>
+<td> BETWEEN </td>
+<td> BTREE </td>
+<td> BY </td>
+<td> CASE </td>
+<td> CLOSED </td></tr>
+<tr class="b">
+<td> CREATE </td>
+<td> COMPACTION </td>
+<td> COMPACT </td>
+<td> CONNECT </td>
+<td> CORRELATE </td>
+<td> DATASET </td></tr>
+<tr class="a">
+<td> COLLECTION </td>
+<td> DATAVERSE </td>
+<td> DECLARE </td>
+<td> DEFINITION </td>
+<td> DECLARE </td>
+<td> DEFINITION </td></tr>
+<tr class="b">
+<td> DELETE </td>
+<td> DESC </td>
+<td> DISCONNECT </td>
+<td> DISTINCT </td>
+<td> DROP </td>
+<td> ELEMENT </td></tr>
+<tr class="a">
+<td> ELEMENT </td>
+<td> EXPLAIN </td>
+<td> ELSE </td>
+<td> ENFORCED </td>
+<td> END </td>
+<td> EVERY </td></tr>
+<tr class="b">
+<td> EXCEPT </td>
+<td> EXIST </td>
+<td> EXTERNAL </td>
+<td> FEED </td>
+<td> FILTER </td>
+<td> FLATTEN </td></tr>
+<tr class="a">
+<td> FOR </td>
+<td> FROM </td>
+<td> FULL </td>
+<td> FUNCTION </td>
+<td> GROUP </td>
+<td> HAVING </td></tr>
+<tr class="b">
+<td> HINTS </td>
+<td> IF </td>
+<td> INTO </td>
+<td> IN </td>
+<td> INDEX </td>
+<td> INGESTION </td></tr>
+<tr class="a">
+<td> INNER </td>
+<td> INSERT </td>
+<td> INTERNAL </td>
+<td> INTERSECT </td>
+<td> IS </td>
+<td> JOIN </td></tr>
+<tr class="b">
+<td> KEYWORD </td>
+<td> LEFT </td>
+<td> LETTING </td>
+<td> LET </td>
+<td> LIKE </td>
+<td> LIMIT </td></tr>
+<tr class="a">
+<td> LOAD </td>
+<td> NODEGROUP </td>
+<td> NGRAM </td>
+<td> NOT </td>
+<td> OFFSET </td>
+<td> ON </td></tr>
+<tr class="b">
+<td> OPEN </td>
+<td> OR </td>
+<td> ORDER </td>
+<td> OUTER </td>
+<td> OUTPUT </td>
+<td> OVER </td></tr>
+<tr class="a">
+<td> PATH </td>
+<td> POLICY </td>
+<td> PRE-SORTED </td>
+<td> PRIMARY </td>
+<td> RAW </td>
+<td> REFRESH </td></tr>
+<tr class="b">
+<td> RETURN </td>
+<td> RTREE </td>
+<td> RUN </td>
+<td> SATISFIES </td>
+<td> SECONDARY </td>
+<td> SELECT </td></tr>
+<tr class="a">
+<td> SET </td>
+<td> SOME </td>
+<td> TEMPORARY </td>
+<td> THEN </td>
+<td> TYPE </td>
+<td> UNKNOWN </td></tr>
+<tr class="b">
+<td> UNNEST </td>
+<td> UPDATE </td>
+<td> USE </td>
+<td> USING </td>
+<td> VALUE </td>
+<td> WHEN </td></tr>
+<tr class="a">
+<td> WHERE </td>
+<td> WITH </td>
+<td> WRITE </td>
+<td>     </td>
+<td>     </td>
+<td>     </td></tr>
+</tbody>
+</table><!--
+ ! 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.
+ !-->
+</div></div></div></div>
+<div class="section">
+<h2><a name="Appendix_2._Performance_Tuning"></a><a name="Performance_tuning" id="Performance_tuning">Appendix 2. Performance Tuning</a></h2><!--
+ ! 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.
+ !-->
+
+<p>The SET statement can be used to override some cluster-wide configuration parameters for a specific request:</p>
+
+<div>
+<div>
+<pre class="source">SET &lt;IDENTIFIER&gt; &lt;STRING_LITERAL&gt;
+</pre></div></div>
+
+<p>As parameter identifiers are qualified names (containing a &#x2018;.&#x2019;) they have to be escaped using backticks (``). Note that changing query parameters will not affect query correctness but only impact performance characteristics, such as response time and throughput.</p></div>
+<div class="section">
+<h2><a name="Parallelism_Parameter"></a><a name="Parallelism_parameter" id="Parallelism_parameter">Parallelism Parameter</a></h2>
+<p>The system can execute each request using multiple cores on multiple machines (a.k.a., partitioned parallelism) in a cluster. A user can manually specify the maximum execution parallelism for a request to scale it up and down using the following parameter:</p>
+<ul>
+
+<li><b>compiler.parallelism</b>: the maximum number of CPU cores can be used to process a query. There are three cases of the value <i>p</i> for compiler.parallelism:
+<ul>
+
+<li>
+
+<p><i>p</i> &lt; 0 or <i>p</i> &gt; the total number of cores in a cluster:  the system will use all available cores in the cluster;</p>
+</li>
+<li>
+
+<p><i>p</i> = 0 (the default):  the system will use the storage parallelism (the number of partitions of stored datasets) as the maximum parallelism for query processing;</p>
+</li>
+<li>
+
+<p>all other cases:  the system will use the user-specified number as the maximum number of CPU cores to use for executing the query.</p>
+</li>
+</ul>
+</li>
+</ul>
+<div class="section">
+<div class="section">
+<div class="section">
+<h5><a name="Example"></a>Example</h5>
+
+<div>
+<div>
+<pre class="source">SET `compiler.parallelism` &quot;16&quot;;
+
+SELECT u.name AS uname, m.message AS message
+FROM GleambookUsers u JOIN GleambookMessages m ON m.authorId = u.id;
+</pre></div></div>
+</div></div></div></div>
+<div class="section">
+<h2><a name="Memory_Parameters"></a><a name="Memory_parameters" id="Memory_parameters">Memory Parameters</a></h2>
+<p>In the system, each blocking runtime operator such as join, group-by and order-by works within a fixed memory budget, and can gracefully spill to disks if the memory budget is smaller than the amount of data they have to hold. A user can manually configure the memory budget of those operators within a query. The supported configurable memory parameters are:</p>
+<ul>
+
+<li>
+
+<p><b>compiler.groupmemory</b>: the memory budget that each parallel group-by operator instance can use; 32MB is the default budget.</p>
+</li>
+<li>
+
+<p><b>compiler.sortmemory</b>: the memory budget that each parallel sort operator instance can use; 32MB is the default budget.</p>
+</li>
+<li>
+
+<p><b>compiler.joinmemory</b>: the memory budget that each parallel hash join operator instance can use; 32MB is the default budget.</p>
+</li>
+<li>
+
+<p><b>compiler.windowmemory</b>: the memory budget that each parallel window aggregate operator instance can use; 32MB is the default budget.</p>
+</li>
+</ul>
+<p>For each memory budget value, you can use a 64-bit integer value with a 1024-based binary unit suffix (for example, B, KB, MB, GB). If there is no user-provided suffix, &#x201c;B&#x201d; is the default suffix. See the following examples.</p>
+<div class="section">
+<div class="section">
+<div class="section">
+<h5><a name="Example"></a>Example</h5>
+
+<div>
+<div>
+<pre class="source">SET `compiler.groupmemory` &quot;64MB&quot;;
+
+SELECT msg.authorId, COUNT(*)
+FROM GleambookMessages msg
+GROUP BY msg.authorId;
+</pre></div></div>
+</div>
+<div class="section">
+<h5><a name="Example"></a>Example</h5>
+
+<div>
+<div>
+<pre class="source">SET `compiler.sortmemory` &quot;67108864&quot;;
+
+SELECT VALUE user
+FROM GleambookUsers AS user
+ORDER BY ARRAY_LENGTH(user.friendIds) DESC;
+</pre></div></div>
+</div>
+<div class="section">
+<h5><a name="Example"></a>Example</h5>
+
+<div>
+<div>
+<pre class="source">SET `compiler.joinmemory` &quot;132000KB&quot;;
+
+SELECT u.name AS uname, m.message AS message
+FROM GleambookUsers u JOIN GleambookMessages m ON m.authorId = u.id;
+</pre></div></div>
+<!--
+ ! 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.
+ !-->
+</div></div></div></div>
+<div class="section">
+<h2><a name="Parallel_Sort_Parameter"></a><a name="Parallel_sort_parameter" id="Parallel_sort_parameter">Parallel Sort Parameter</a></h2>
+<p>The following parameter enables you to activate or deactivate full parallel sort for order-by operations.</p>
+<p>When full parallel sort is inactive (<tt>false</tt>), each existing data partition is sorted (in parallel), and then all data partitions are merged into a single node.</p>
+<p>When full parallel sort is active (<tt>true</tt>), the data is first sampled, and then repartitioned so that each partition contains data that is greater than the previous partition. The data in each partition is then sorted (in parallel), but the sorted partitions are not merged into a single node.</p>
+<ul>
+
+<li><b>compiler.sort.parallel</b>: A boolean specifying whether full parallel sort is active (<tt>true</tt>) or inactive (<tt>false</tt>). The default value is <tt>true</tt>.</li>
+</ul>
+<div class="section">
+<div class="section">
+<div class="section">
+<h5><a name="Example"></a>Example</h5>
+
+<div>
+<div>
+<pre class="source">SET `compiler.sort.parallel` &quot;true&quot;;
+
+SELECT VALUE user
+FROM GleambookUsers AS user
+ORDER BY ARRAY_LENGTH(user.friendIds) DESC;
+</pre></div></div>
+<!--
+ ! 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.
+ !-->
+</div></div></div></div>
+<div class="section">
+<h2><a name="Controlling_Index-Only-Plan_Parameter"></a><a name="Index_Only" id="Index_Only">Controlling Index-Only-Plan Parameter</a></h2>
+<p>By default, the system tries to build an index-only plan whenever utilizing a secondary index is possible. For example, if a SELECT or JOIN query can utilize an enforced B+Tree or R-Tree index on a field, the optimizer checks whether a secondary-index search alone can generate the result that the query asks for. It mainly checks two conditions: (1) predicates used in WHERE only uses the primary key field and/or secondary key field and (2) the result does not return any other fields. If these two conditions hold, it builds an index-only plan. Since an index-only plan only searches a secondary-index to answer a query, it is faster than a non-index-only plan that needs to search the primary index. However, this index-only plan can be turned off per query by setting the following parameter.</p>
+<ul>
+
+<li><b>compiler.indexonly</b>: if this is set to false, the index-only-plan will not be applied; the default value is true.</li>
+</ul>
+<div class="section">
+<div class="section">
+<div class="section">
+<h5><a name="Example"></a>Example</h5>
+
+<div>
+<div>
+<pre class="source">set `compiler.indexonly` &quot;false&quot;;
+
+SELECT m.message AS message
+FROM GleambookMessages m where m.message = &quot; love product-b its shortcut-menu is awesome:)&quot;;
+</pre></div></div>
+<!--
+ ! 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.
+ !-->
+</div></div></div></div>
+<div class="section">
+<h2><a name="Appendix_3._Variable_Bindings_and_Name_Resolution"></a><a name="Variable_bindings_and_name_resolution" id="Variable_bindings_and_name_resolution">Appendix 3. Variable Bindings and Name Resolution</a></h2><!--
+ ! 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.
+ !-->
+
+<p>In this Appendix, we&#x2019;ll look at how variables are bound and how names are resolved. Names can appear in every clause of a query. Sometimes a name consists of just a single identifier, e.g., <tt>region</tt> or <tt>revenue</tt>. More often a name will consist of two identifiers separated by a dot, e.g., <tt>customer.address</tt>. Occasionally a name may have more than two identifiers, e.g., <tt>policy.owner.address.zipcode</tt>. <i>Resolving</i> a name means determining exactly what the (possibly multi-part) name refers to. It is necessary to have well-defined rules for how to resolve a name in cases of ambiguity. (In the absence of schemas, such cases arise more commonly, and also differently, than they do in SQL.)</p>
+<p>The basic job of each clause in a query block is to bind variables. Each clause sees the variables bound by previous clauses and may bind additional variables. Names are always resolved with respect to the variables that are bound (&#x201c;in scope&#x201d;) at the place where the name use in question occurs. It is possible that the name resolution process will fail, which may lead to an empty result or an error message.</p>
+<p>One important bit of background: When the system is reading a query and resolving its names, it has a list of all the available dataverses and datasets. As a result, it knows whether <tt>a.b</tt> is a valid name for dataset <tt>b</tt> in dataverse <tt>a</tt>. However, the system does not in general have knowledge of the schemas of the data inside the datasets; remember that this is a much more open world. As a result, in general the system cannot know whether any object in a particular dataset will have a field named <tt>c</tt>. These assumptions affect how errors are handled. If you try to access dataset <tt>a.b</tt> and no dataset by that name exists, you will get an error and your query will not run. However, if you try to access a field <tt>c</tt> in a collection of objects, your query will run and return <tt>missing</tt> for each object that doesn&#x2019;t have a field named <tt>c</tt> &#x2013; this is because it&#x2019;s possible that some object (someday) could have such a field.</p></div>
+<div class="section">
+<h2><a name="Binding_Variables"></a><a name="Binding_variables" id="Binding_variables">Binding Variables</a></h2>
+<p>Variables can be bound in the following ways:</p>
+<ol style="list-style-type: decimal">
+
+<li>
+
+<p>WITH and LET clauses bind a variable to the result of an expression in a straightforward way</p>
+<p>Examples:</p>
+<p><tt>WITH cheap_parts AS (SELECT partno FROM parts WHERE price &lt; 100)</tt> binds the variable <tt>cheap_parts</tt> to the result of the subquery.</p>
+<p><tt>LET pay = salary + bonus</tt> binds the variable <tt>pay</tt> to the result of evaluating the expression <tt>salary + bonus</tt>.</p>
+</li>
+<li>
+
+<p>FROM, GROUP BY, and SELECT clauses have optional AS subclauses that contain an expression and a name (called an <i>iteration variable</i> in a FROM clause, or an alias in GROUP BY or SELECT.)</p>
+<p>Examples:</p>
+<p><tt>FROM customer AS c, order AS o</tt></p>
+<p><tt>GROUP BY salary + bonus AS total_pay</tt></p>
+<p><tt>SELECT MAX(price) AS highest_price</tt></p>
+<p>An AS subclause always binds the name (as a variable) to the result of the expression (or, in the case of a FROM clause, to the <i>individual members</i> of the collection identified by the expression.)</p>
+<p>It&#x2019;s always a good practice to use the keyword AS when defining an alias or iteration variable. However, as in SQL, the syntax allows the keyword AS to be omitted. For example, the FROM clause above could have been written like this:</p>
+<p><tt>FROM customer c, order o</tt></p>
+<p>Omitting the keyword AS does not affect the binding of variables. The FROM clause in this example binds variables c and o whether the keyword AS is used or not.</p>
+<p>In certain cases, a variable is automatically bound even if no alias or variable-name is specified. Whenever an expression could have been followed by an AS subclause, if the expression consists of a simple name or a path expression, that expression binds a variable whose name is the same as the simple name or the last step in the path expression. Here are some examples:</p>
+<p><tt>FROM customer, order</tt> binds iteration variables named <tt>customer</tt> and <tt>order</tt></p>
+<p><tt>GROUP BY address.zipcode</tt> binds a variable named <tt>zipcode</tt></p>
+<p><tt>SELECT item[0].price</tt> binds a variable named <tt>price</tt></p>
+<p>Note that a FROM clause iterates over a collection (usually a dataset), binding a variable to each member of the collection in turn. The name of the collection remains in scope, but it is not a variable. For example, consider this FROM clause used in a self-join:</p>
+<p><tt>FROM customer AS c1, customer AS c2</tt></p>
+<p>This FROM clause joins the customer dataset to itself, binding the iteration variables c1 and c2 to objects in the left-hand-side and right-hand-side of the join, respectively. After the FROM clause, c1 and c2 are in scope as variables, and customer remains accessible as a dataset name but not as a variable.</p>
+</li>
+<li>
+
+<p>Special rules for GROUP BY:</p>
+<ol style="list-style-type: decimal">
+
+<li>
+
+<p>If a GROUP BY clause specifies an expression that has no explicit alias, it binds a pseudo-variable that is lexicographically identical to the expression itself. For example:</p>
+<p><tt>GROUP BY salary + bonus</tt> binds a pseudo-variable named <tt>salary + bonus</tt>.</p>
+<p>This rule allows subsequent clauses to refer to the grouping expression (salary + bonus) even though its constituent variables (salary and bonus) are no longer in scope. For example, the following query is valid:</p>
+
+<div>
+<div>
+<pre class="source">FROM employee
+GROUP BY salary + bonus
+HAVING salary + bonus &gt; 1000
+SELECT salary + bonus, COUNT(*) AS how_many
+</pre></div></div>
+
+<p>While it might have been more elegant to explicitly require an alias in cases like this, the pseudo-variable rule is retained for SQL compatibility. Note that the expression <tt>salary + bonus</tt> is not <i>actually</i> evaluated in the HAVING and SELECT clauses (and could not be since <tt>salary</tt> and <tt>bonus</tt> are no longer individually in scope). Instead, the expression <tt>salary + bonus</tt> is treated as a reference to the pseudo-variable defined in the GROUP BY clause.</p>
+</li>
+<li>
+
+<p>A GROUP BY clause may be followed by a GROUP AS clause that binds a variable to the group. The purpose of this variable is to make the individual objects inside the group visible to subqueries that may need to iterate over them.</p>
+<p>The GROUP AS variable is bound to a multiset of objects. Each object represents one of the members of the group. Since the group may have been formed from a join, each of the member-objects contains a nested object for each variable bound by the nearest FROM clause (and its LET subclause, if any). These nested objects, in turn, contain the actual fields of the group-member. To understand this process, consider the following query fragment:</p>
+
+<div>
+<div>
+<pre class="source">FROM parts AS p, suppliers AS s
+WHERE p.suppno = s.suppno
+GROUP BY p.color GROUP AS g
+</pre></div></div>
+
+<p>Suppose that the objects in <tt>parts</tt> have fields <tt>partno</tt>, <tt>color</tt>, and <tt>suppno</tt>. Suppose that the objects in suppliers have fields <tt>suppno</tt> and <tt>location</tt>.</p>
+<p>Then, for each group formed by the GROUP BY, the variable g will be bound to a multiset with the following structure:</p>
+
+<div>
+<div>
+<pre class="source">[ { &quot;p&quot;: { &quot;partno&quot;: &quot;p1&quot;, &quot;color&quot;: &quot;red&quot;, &quot;suppno&quot;: &quot;s1&quot; },
+    &quot;s&quot;: { &quot;suppno&quot;: &quot;s1&quot;, &quot;location&quot;: &quot;Denver&quot; } },
+  { &quot;p&quot;: { &quot;partno&quot;: &quot;p2&quot;, &quot;color&quot;: &quot;red&quot;, &quot;suppno&quot;: &quot;s2&quot; },
+    &quot;s&quot;: { &quot;suppno&quot;: &quot;s2&quot;, &quot;location&quot;: &quot;Atlanta&quot; } },
+  ...
+]
+</pre></div></div>
+</li>
+</ol>
+</li>
+</ol></div>
+<div class="section">
+<h2><a name="Scoping" id="Scoping">Scoping</a></h2>
+<p>In general, the variables that are in scope at a particular position are those variables that were bound earlier in the current query block, in outer (enclosing) query blocks, or in a WITH clause at the beginning of the query. More specific rules follow.</p>
+<p>The clauses in a query block are conceptually processed in the following order:</p>
+<ul>
+
+<li>FROM (followed by LET subclause, if any)</li>
+<li>WHERE</li>
+<li>GROUP BY (followed by LET subclause, if any)</li>
+<li>HAVING</li>
+<li>SELECT or SELECT VALUE</li>
+<li>ORDER BY</li>
+<li>OFFSET</li>
+<li>LIMIT</li>
+</ul>
+<p>During processing of each clause, the variables that are in scope are those variables that are bound in the following places:</p>
+<ol style="list-style-type: decimal">
+
+<li>
+
+<p>In earlier clauses of the same query block (as defined by the ordering given above).</p>
+<p>Example: <tt>FROM orders AS o SELECT o.date</tt> The variable <tt>o</tt> in the SELECT clause is bound, in turn, to each object in the dataset <tt>orders</tt>.</p>
+</li>
+<li>
+
+<p>In outer query blocks in which the current query block is nested. In case of duplication, the innermost binding wins.</p>
+</li>
+<li>
+
+<p>In the WITH clause (if any) at the beginning of the query.</p>
+</li>
+</ol>
+<p>However, in a query block where a GROUP BY clause is present:</p>
+<ol style="list-style-type: decimal">
+
+<li>
+
+<p>In clauses processed before GROUP BY, scoping rules are the same as though no GROUP BY were present.</p>
+</li>
+<li>
+
+<p>In clauses processed after GROUP BY, the variables bound in the nearest FROM-clause (and its LET subclause, if any) are removed from scope and replaced by the variables bound in the GROUP BY clause (and its LET subclause, if any). However, this replacement does not apply inside the arguments of the five SQL special aggregating functions (MIN, MAX, AVG, SUM, and COUNT). These functions still need to see the individual data items over which they are computing an aggregation. For example, after <tt>FROM employee AS e GROUP BY deptno</tt>, it would not be valid to reference <tt>e.salary</tt>, but <tt>AVG(e.salary)</tt> would be valid.</p>
+</li>
+</ol>
+<p>Special case: In an expression inside a FROM clause, a variable is in scope if it was bound in an earlier expression in the same FROM clause. Example:</p>
+
+<div>
+<div>
+<pre class="source">FROM orders AS o, o.items AS i
+</pre></div></div>
+
+<p>The reason for this special case is to support iteration over nested collections.</p>
+<p>Note that, since the SELECT clause comes <i>after</i> the WHERE and GROUP BY clauses in conceptual processing order, any variables defined in SELECT are not visible in WHERE or GROUP BY. Therefore the following query will not return what might be the expected result (since in the WHERE clause, <tt>pay</tt> will be interpreted as a field in the <tt>emp</tt> object rather than as the computed value <tt>salary + bonus</tt>):</p>
+
+<div>
+<div>
+<pre class="source">SELECT name, salary + bonus AS pay
+FROM emp
+WHERE pay &gt; 1000
+ORDER BY pay
+</pre></div></div>
+
+<p>The likely intent of the query above can be accomplished as follows:</p>
+
+<div>
+<div>
+<pre class="source">FROM emp AS e
+LET pay = e.salary + e.bonus
+WHERE pay &gt; 1000
+SELECT e.name, pay
+ORDER BY pay
+</pre></div></div>
+
+<p>Note that variables defined by <tt>JOIN</tt> subclauses are not visible to other subclauses in the same <tt>FROM</tt> clause. This also applies to the <tt>FROM</tt> variable that starts the <tt>JOIN</tt> subclause.</p></div>
+<div class="section">
+<h2><a name="Resolving_Names"></a><a name="Resolving_names" id="Resolving_names">Resolving Names</a></h2>
+<p>The process of name resolution begins with the leftmost identifier in the name. The rules for resolving the leftmost identifier are:</p>
+<ol style="list-style-type: decimal">
+
+<li>
+
+<p><i>In a FROM clause</i>: Names in a FROM clause identify the collections over which the query block will iterate. These collections may be stored datasets or may be the results of nested query blocks. A stored dataset may be in a named dataverse or in the default dataverse. Thus, if the two-part name <tt>a.b</tt> is in a FROM clause, a might represent a dataverse and <tt>b</tt> might represent a dataset in that dataverse. Another example of a two-part name in a FROM clause is <tt>FROM orders AS o, o.items AS i</tt>. In <tt>o.items</tt>, <tt>o</tt> represents an order object bound earlier in the FROM clause, and items represents the items object inside that order.</p>
+<p>The rules for resolving the leftmost identifier in a FROM clause (including a JOIN subclause), or in the expression following IN in a quantified predicate, are as follows:</p>
+<ol style="list-style-type: decimal">
+
+<li>
+
+<p>If the identifier matches a variable-name that is in scope, it resolves to the binding of that variable. (Note that in the case of a subquery, an in-scope variable might have been bound in an outer query block; this is called a correlated subquery.)</p>
+</li>
+<li>
+
+<p>Otherwise, if the identifier is the first part of a two-part name like <tt>a.b</tt>, the name is treated as <tt>dataverse.dataset</tt>. If the identifier stands alone as a one-part name, it is treated as the name of a dataset in the default dataverse. If the designated dataset exists then the identifier is resolved to that dataset, otherwise if a synonym with given name exists then the identifier is resolved to the target dataset of that synonym (potentially recursively if this synonym points to another synonym). An error will result if the designated dataset or a synonym with this name does not exist.</p>
+<p>Datasets take precedence over synonyms, so if both a dataset and a synonym have the same name then the resolution is to the dataset.</p>
+</li>
+</ol>
+</li>
+<li>
+
+<p><i>Elsewhere in a query block</i>: In clauses other than FROM, a name typically identifies a field of some object. For example, if the expression <tt>a.b</tt> is in a SELECT or WHERE clause, it&#x2019;s likely that <tt>a</tt> represents an object and <tt>b</tt> represents a field in that object.</p>
+<p>The rules for resolving the leftmost identifier in clauses other than the ones listed in Rule 1 are:</p>
+<ol style="list-style-type: decimal">
+
+<li>
+
+<p>If the identifier matches a variable-name that is in scope, it resolves to the binding of that variable. (In the case of a correlated subquery, the in-scope variable might have been bound in an outer query block.)</p>
+</li>
+<li>
+
+<p>(The &#x201c;Single Variable Rule&#x201d;): Otherwise, if the FROM clause in the current query block binds exactly one variable, the identifier is treated as a field access on the object bound to that variable. For example, in the query <tt>FROM customer SELECT address</tt>, the identifier address is treated as a field in the object bound to the variable customer. At runtime, if the object bound to customer has no <tt>address</tt> field, the <tt>address</tt> expression will return <tt>missing</tt>. If the FROM clause in the current query block binds multiple variables, name resolution fails with an &#x201c;ambiguous name&#x201d; error. If there&#x2019;s no FROM clause in the current query block, name resolution fails with an &#x201c;undefined identifier&#x201d; error. Note that the Single Variable Rule searches for bound variables only in the current query block, not in outer (containing) blocks. The purpose of this rule is to permit the compiler to resolve field-references unambiguously without relying on any schema information. Also note that variables defined by LET clauses do not participate in the resolution process performed by this rule.</p>
+<p>Exception: In a query that has a GROUP BY clause, the Single Variable Rule does not apply in any clauses that occur after the GROUP BY because, in these clauses, the variables bound by the FROM clause are no longer in scope. In clauses after GROUP BY, only Rule 2.1 applies.</p>
+</li>
+</ol>
+</li>
+<li>
+
+<p>In an ORDER BY clause following a UNION ALL expression:</p>
+<p>The leftmost identifier is treated as a field-access on the objects that are generated by the UNION ALL. For example:</p>
+
+<div>
+<div>
+<pre class="source">query-block-1
+UNION ALL
+query-block-2
+ORDER BY salary
+</pre></div></div>
+
+<p>In the result of this query, objects that have a foo field will be ordered by the value of this field; objects that have no foo field will appear at at the beginning of the query result (in ascending order) or at the end (in descending order.)</p>
+</li>
+<li>
+
+<p><i>In a standalone expression</i>: If a query consists of a standalone expression then identifiers inside that expression are resolved according to Rule 1. For example, if the whole query is <tt>ARRAY_COUNT(a.b)</tt> then <tt>a.b</tt> will be treated as dataset <tt>b</tt> contained in dataverse <tt>a</tt>. Note that this rule only applies to identifiers which are located directly inside a standalone expression. Identifiers inside SELECT statements in a standalone expression are still resolved according to Rules 1-3. For example, if the whole query is <tt>ARRAY_SUM( (FROM employee AS e SELECT VALUE salary) )</tt> then <tt>salary</tt> is resolved as <tt>e.salary</tt> following the &#x201c;Single Variable Rule&#x201d; (Rule 2.2).</p>
+</li>
+<li>
+
+<p>Once the leftmost identifier has been resolved, the following dots and identifiers in the name (if any) are treated as a path expression that navigates to a field nested inside that object. The name resolves to the field at the end of the path. If this field does not exist, the value <tt>missing</tt> is returned.</p>
+</li>
+</ol></div>
+        </div>
+      </div>
+    </div>
+    <hr/>
+    <footer>
+      <div class="container-fluid">
+        <div class="row-fluid">
+<div class="row-fluid">Apache AsterixDB, AsterixDB, Apache, the Apache
+        feather logo, and the Apache AsterixDB project logo are either
+        registered trademarks or trademarks of The Apache Software
+        Foundation in the United States and other countries.
+        All other marks mentioned may be trademarks or registered
+        trademarks of their respective owners.
+      </div>
+        </div>
+      </div>
+    </footer>
+  </body>
+</html>