Add note that REINDEX takes a weaker lock than it used to.
authorTom Lane <tgl@sss.pgh.pa.us>
Sun, 24 Oct 2004 22:43:56 +0000 (22:43 +0000)
committerTom Lane <tgl@sss.pgh.pa.us>
Sun, 24 Oct 2004 22:43:56 +0000 (22:43 +0000)
doc/src/sgml/release.sgml

index 0ca053c27d5c96bf9339b3e69d9fdc174c9d9f52..466754b11ffd70b93c3ec39d8922fc9a6b84a22f 100644 (file)
@@ -1,5 +1,5 @@
 <!--
-$PostgreSQL: pgsql/doc/src/sgml/release.sgml,v 1.302 2004/10/22 00:25:18 tgl Exp $
+$PostgreSQL: pgsql/doc/src/sgml/release.sgml,v 1.303 2004/10/24 22:43:56 tgl Exp $
 -->
 
 <appendix id="release">
@@ -1170,6 +1170,18 @@ $PostgreSQL: pgsql/doc/src/sgml/release.sgml,v 1.302 2004/10/22 00:25:18 tgl Exp
      </para>
     </listitem>
 
+    <listitem>
+     <para>
+      <command>REINDEX</command> does not exclusively lock the index's
+      parent table anymore
+     </para>
+     <para>
+      The index itself is still exclusively locked, but readers of the
+      table can continue if they are not using the particular index
+      being rebuilt.
+     </para>
+    </listitem>
+
     <listitem>
      <para>
       Erase MD5 user passwords when a user is renamed (Bruce)