Return FALSE instead of throwing error for comparisons with empty ranges.

Change range_before, range_after, range_adjacent to return false rather
than throwing an error when one or both input ranges are empty.

The original definition is unnecessarily difficult to use, and also can
result in undesirable planner failures since the planner could try to
compare an empty range to something else while deriving statistical
estimates.  (This was, in fact, the cause of repeatable regression test
failures on buildfarm member jaguar, as well as intermittent failures
elsewhere.)

Also tweak rangetypes regression test to not drop all the objects it
creates, so that the final state of the regression database contains
some rangetype objects for pg_dump testing.

2 files changed, 44 insertions, 31 deletions

diff --git a/doc/src/sgml/filelist.sgml b/doc/src/sgml/filelist.sgml
index fb69415f800..7a698e59b3e 100644
--- a/doc/src/sgml/filelist.sgml
+++ b/doc/src/sgml/filelist.sgml
@@ -25,7 +25,7 @@
 <!ENTITY mvcc       SYSTEM "mvcc.sgml">
 <!ENTITY perform    SYSTEM "perform.sgml">
 <!ENTITY queries    SYSTEM "queries.sgml">
-<!entity rangetypes SYSTEM "rangetypes.sgml">
+<!ENTITY rangetypes SYSTEM "rangetypes.sgml">
 <!ENTITY rowtypes   SYSTEM "rowtypes.sgml">
 <!ENTITY syntax     SYSTEM "syntax.sgml">
 <!ENTITY textsearch SYSTEM "textsearch.sgml">
diff --git a/doc/src/sgml/func.sgml b/doc/src/sgml/func.sgml
index 7ecf7510435..b9ef827984d 100644
--- a/doc/src/sgml/func.sgml
+++ b/doc/src/sgml/func.sgml
@@ -10461,6 +10461,10 @@ SELECT NULLIF(value, '(none)') ...
   <title>Range Functions and Operators</title>
 
   <para>
+   See <xref linkend="rangetypes"> for an overview of range types.
+  </para>
+
+  <para>
    <xref linkend="range-operators-table"> shows the operators
    available for range types.
   </para>
@@ -10556,65 +10560,74 @@ SELECT NULLIF(value, '(none)') ...
 
        <row>
         <entry> <literal>&amp;&lt;</literal> </entry>
-        <entry>Does not extend to the right of?</entry>
+        <entry>does not extend to the right of</entry>
         <entry><literal>int8range(1,20) &amp;&lt; int8range(18,20)</literal></entry>
         <entry><literal>t</literal></entry>
        </row>
 
        <row>
         <entry> <literal>&amp;&gt;</literal> </entry>
-        <entry>Does not extend to the left of?</entry>
+        <entry>does not extend to the left of</entry>
         <entry><literal>int8range(7,20) &amp;&gt; int8range(5,10)</literal></entry>
         <entry><literal>t</literal></entry>
        </row>
 
        <row>
         <entry> <literal>-|-</literal> </entry>
-        <entry>adjacent?</entry>
+        <entry>is adjacent to</entry>
         <entry><literal>numrange(1.1,2.2) -|- numrange(2.2,3.3)</literal></entry>
         <entry><literal>t</literal></entry>
        </row>
 
        <row>
         <entry> <literal>+</literal> </entry>
-        <entry>Union</entry>
+        <entry>union</entry>
         <entry><literal>numrange(5,15) + numrange(10,20)</literal></entry>
         <entry><literal>[5,20)</literal></entry>
        </row>
 
        <row>
-        <entry> <literal>-</literal> </entry>
-        <entry>Difference</entry>
-        <entry><literal>int8range(5,15) - int8range(10,20)</literal></entry>
-        <entry><literal>[5,10)</literal></entry>
-       </row>
-
-       <row>
         <entry> <literal>*</literal> </entry>
-        <entry>Intersection</entry>
+        <entry>intersection</entry>
         <entry><literal>int8range(5,15) * int8range(10,20)</literal></entry>
         <entry><literal>[10,15)</literal></entry>
        </row>
 
+       <row>
+        <entry> <literal>-</literal> </entry>
+        <entry>difference</entry>
+        <entry><literal>int8range(5,15) - int8range(10,20)</literal></entry>
+        <entry><literal>[5,10)</literal></entry>
+       </row>
+
       </tbody>
      </tgroup>
     </table>
 
   <para>
-   Range comparisons compare the lower bounds first, and only if
-   equal, compare the upper bounds. This is generally most useful for
-   B-tree indexes, rather than being useful comparisons by themselves.
+   The simple comparison operators <literal>&lt;</literal>,
+   <literal>&gt;</literal>, <literal>&lt;=</literal>, and
+   <literal>&gt;=</literal> compare the lower bounds first, and only if those
+   are equal, compare the upper bounds.  These comparisons are not usually
+   very useful for ranges, but are provided to allow B-tree indexes to be
+   constructed on ranges.
   </para>
 
   <para>
-   See <xref linkend="rangetypes"> for more details about range operator
-   behavior.
+   The left-of/right-of/adjacent operators always return false when an empty
+   range is involved; that is, an empty range is not considered to be either
+   before or after any other range.
+  </para>
+
+  <para>
+   The union and difference operators will fail if the resulting range would
+   need to contain two disjoint sub-ranges, as such a range cannot be
+   represented.
   </para>
 
   <para>
    <xref linkend="range-functions-table"> shows the functions
-   available for use with range types. See <xref linkend="rangetypes">
-   for more information  and examples of the use of these functions.
+   available for use with range types.
   </para>
 
   <indexterm>
@@ -10658,7 +10671,7 @@ SELECT NULLIF(value, '(none)') ...
           <function>lower</function>(<type>anyrange</type>)
          </literal>
         </entry>
-        <entry><type>anyrange</type></entry>
+        <entry>range's element type</entry>
         <entry>lower bound of range</entry>
         <entry><literal>lower(numrange(1.1,2.2))</literal></entry>
         <entry><literal>1.1</literal></entry>
@@ -10669,7 +10682,7 @@ SELECT NULLIF(value, '(none)') ...
           <function>upper</function>(<type>anyrange</type>)
          </literal>
         </entry>
-        <entry><type>anyrange</type></entry>
+        <entry>range's element type</entry>
         <entry>upper bound of range</entry>
         <entry><literal>upper(numrange(1.1,2.2))</literal></entry>
         <entry><literal>2.2</literal></entry>
@@ -10680,7 +10693,7 @@ SELECT NULLIF(value, '(none)') ...
           <function>isempty</function>(<type>anyrange</type>)
          </literal>
         </entry>
-        <entry><type>anyrange</type></entry>
+        <entry><type>boolean</type></entry>
         <entry>is the range empty?</entry>
         <entry><literal>isempty(numrange(1.1,2.2))</literal></entry>
         <entry><literal>false</literal></entry>
@@ -10691,8 +10704,8 @@ SELECT NULLIF(value, '(none)') ...
           <function>lower_inc</function>(<type>anyrange</type>)
          </literal>
         </entry>
-        <entry><type>anyrange</type></entry>
-        <entry>is the lower bound of the range inclusive?</entry>
+        <entry><type>boolean</type></entry>
+        <entry>is the lower bound inclusive?</entry>
         <entry><literal>lower_inc(numrange(1.1,2.2))</literal></entry>
         <entry><literal>true</literal></entry>
        </row>
@@ -10702,8 +10715,8 @@ SELECT NULLIF(value, '(none)') ...
           <function>upper_inc</function>(<type>anyrange</type>)
          </literal>
         </entry>
-        <entry><type>anyrange</type></entry>
-        <entry>is the upper bound of the range inclusive?</entry>
+        <entry><type>boolean</type></entry>
+        <entry>is the upper bound inclusive?</entry>
         <entry><literal>upper_inc(numrange(1.1,2.2))</literal></entry>
         <entry><literal>false</literal></entry>
        </row>
@@ -10713,8 +10726,8 @@ SELECT NULLIF(value, '(none)') ...
           <function>lower_inf</function>(<type>anyrange</type>)
          </literal>
         </entry>
-        <entry><type>anyrange</type></entry>
-        <entry>is the lower bound of the range infinite?</entry>
+        <entry><type>boolean</type></entry>
+        <entry>is the lower bound infinite?</entry>
         <entry><literal>lower_inf('(,)'::daterange)</literal></entry>
         <entry><literal>true</literal></entry>
        </row>
@@ -10724,8 +10737,8 @@ SELECT NULLIF(value, '(none)') ...
           <function>upper_inf</function>(<type>anyrange</type>)
          </literal>
         </entry>
-        <entry><type>anyrange</type></entry>
-        <entry>is the upper bound of the range infinite?</entry>
+        <entry><type>boolean</type></entry>
+        <entry>is the upper bound infinite?</entry>
         <entry><literal>upper_inf('(,)'::daterange)</literal></entry>
         <entry><literal>true</literal></entry>
        </row>