+Let's suppose that your database have table C<files> with following structure:
+
+ id: int
+ filename: text
+ size: int
+ content: text
+ writable: boolean
+
+Following is example how to mount table like that to C</mnt>:
+
+ my $mnt = Fuse::DBI->mount({
+ 'filenames' => 'select id,filename,size,writable from files',
+ 'read' => 'select content from files where id = ?',
+ 'update' => 'update files set content = ? where id = ?',
+ 'dsn' => 'DBI:Pg:dbname=test_db',
+ 'user' => 'database_user',
+ 'password' => 'database_password',
+ 'invalidate' => sub { ... },
+ });
+
+Options:
+
+=over 5
+
+=item filenames
+
+SQL query which returns C<id> (unique id for that row), C<filename>,
+C<size> and C<writable> boolean flag.
+
+=item read
+
+SQL query which returns only one column with content of file and has
+placeholder C<?> for C<id>.
+
+=item update
+
+SQL query with two pace-holders, one for new content and one for C<id>.
+
+=item dsn
+
+C<DBI> dsn to connect to (contains database driver and name of database).
+
+=item user
+
+User with which to connect to database
+
+=item password
+
+Password for connecting to database
+
+=item invalidate
+
+Optional anonymous code reference which will be executed when data is updated in
+database. It can be used as hook to delete cache (for example on-disk-cache)
+which is created from data edited through C<Fuse::DBI>.
+
+=item fork
+
+Optional flag which forks after mount so that executing script will continue
+running. Implementation is experimental.
+
+=back
+
+There is also alternative way which can generate C<read> and C<update>
+queries on the fly:
+