Depending on the resultset attributes one of:
SELECT * FROM (
- SELECT *, ROWNUM rownum__index FROM (
+ SELECT *, ROWNUM AS rownum__index FROM (
SELECT ...
) WHERE ROWNUM <= ($limit+$offset)
) WHERE rownum__index >= ($offset+1)
or
SELECT * FROM (
- SELECT *, ROWNUM rownum__index FROM (
+ SELECT *, ROWNUM AS rownum__index FROM (
SELECT ...
)
) WHERE rownum__index BETWEEN ($offset+1) AND ($limit+$offset)
return <<EOS;
SELECT $sq_attrs->{selection_outer} FROM (
- SELECT $sq_attrs->{selection_outer}, ROWNUM $idx_name FROM (
+ SELECT $sq_attrs->{selection_outer}, ROWNUM AS $idx_name FROM (
SELECT $sq_attrs->{selection_inner} $sq_attrs->{query_leftover}${order_group_having}
) $qalias WHERE ROWNUM <= ?
) $qalias WHERE $idx_name >= ?
return <<EOS;
SELECT $sq_attrs->{selection_outer} FROM (
- SELECT $sq_attrs->{selection_outer}, ROWNUM $idx_name FROM (
+ SELECT $sq_attrs->{selection_outer}, ROWNUM AS $idx_name FROM (
SELECT $sq_attrs->{selection_inner} $sq_attrs->{query_leftover}${order_group_having}
) $qalias
) $qalias WHERE $idx_name BETWEEN ? AND ?
databases. It works by ordering the set by some unique column, and calculating
the amount of rows that have a less-er value (thus emulating a L</RowNum>-like
index). Of course this implies the set can only be ordered by a single unique
-column. Also note that this technique can be and often is B<excruciatingly
-slow>.
+column.
+
+Also note that this technique can be and often is B<excruciatingly slow>. You
+may have much better luck using L<DBIx::Class::ResultSet/software_limit>
+instead.
Currently used by B<Sybase ASE>, due to lack of any other option.
next if $in_sel_index->{$chunk};
$extra_order_sel->{$chunk} ||= $self->_quote (
- 'ORDER__BY__' . scalar keys %{$extra_order_sel||{}}
+ 'ORDER__BY__' . sprintf '%03d', scalar keys %{$extra_order_sel||{}}
);
}