Bug#30480: Falcon: searches fail if LIKE and key partition

(also fixes the bugs: Bug#29320, Bug#29493 and Bug#30536)

Problem: Partitioning did not handle unordered scans correctly
for engines with unordered read order.

Solution: do not stop scanning fi a recored is out of range, since
there can be more records within the range afterwards.

Note: this is the patch that fixes the bug, but since there are no
storage engines shipped with mysql 5.1 (falcon comes in 6.0) there
are no test cases (it is a separate patch that only goes into 6.0)


sql/ha_partition.cc:
  Bug#30480: Falcon: searches fail if LIKE and key partition
  
  Problem was that partitioning did not handle unordered scans correctly
  for engines with unordered read order.
  
  Solution: do not stop if a recored is out of range, since it can come
  more records within the range afterwards

1 files changed, 4 insertions, 2 deletions

diff --git a/sql/ha_partition.cc b/sql/ha_partition.cc
index 1400d9da753..5abdf873c81 100644
--- a/sql/ha_partition.cc
+++ b/sql/ha_partition.cc
@@ -3983,7 +3983,8 @@ int ha_partition::handle_unordered_next(uchar *buf, bool is_next_same)
   }
   else if (!(error= file->index_next(buf)))
   {
-    if (compare_key(end_range) <= 0)
+    if (!(file->table_flags() & HA_READ_ORDER) ||
+        compare_key(end_range) <= 0)
     {
       m_last_part= m_part_spec.start_part;
       DBUG_RETURN(0);                           // Row was in range
@@ -4060,7 +4061,8 @@ int ha_partition::handle_unordered_scan_next_partition(uchar * buf)
     }
     if (!error)
     {
-      if (compare_key(end_range) <= 0)
+      if (!(file->table_flags() & HA_READ_ORDER) ||
+          compare_key(end_range) <= 0)
       {
         m_last_part= i;
         DBUG_RETURN(0);