@@ -231,23 +231,22 @@ doc/src/sgml/ref/pg_dumpall.sgml
231
231
</listitem>
232
232
</varlistentry>
233
233
234
- <varlistentry>
234
+ <varlistentry>
235
235
<term><option>--add-collprovider</option></term>
236
236
<listitem>
237
237
<para>
238
- This option is recommended when upgrading with
238
+ If the collation provider for the database to be dumped is implicit,
239
+ this option explicitly adds the collation provider to the dump.
240
+ This option is recommended if you are using
239
241
<application>pg_dumpall</application> and
240
- <application>pg_restore</application> from previous versions of
241
- <productname>&project;</> as well as from all versions of
242
- <productname>PostgreSQL</> (it's automatically used in case of a binary
243
- upgrade; and it actually does nothing for the current version of
244
- <productname>&project;</>). If you do not want to manually fix the
245
- scripts (or do not know how to do it correctly) this helps to avoid mess
246
- when the templated database and the created database actually use
247
- different collation providers that are ommitted (see <xref
248
- linkend="collation-managing">). In this case the check constrains that
242
+ <application>pg_restore</application> to upgrade from previous versions of
243
+ <productname>&project;</> or any version of <productname>PostgreSQL</>.
244
+ It allows to preserve the original collation provider for the migrated database if
245
+ the <literal>template0</literal> database in the new cluster uses different
246
+ collation settings. Otherwise, check constraints that
249
247
use the default collation of the database may change and the
250
- <command>COPY</command> command may end with fail.
248
+ <command>COPY</command> command may end with a failure.
249
+ For binary upgrades, this option is used automatically.
251
250
</para>
252
251
</listitem>
253
252
</varlistentry>
0 commit comments