Fix for BUG#35570 "CHECKSUM TABLE unreliable if LINESTRING field (same content / differen checksum)"

This will be back-ported to 5.x trees but the work for R-tree logging critically needs this patch in Maria now.


mysql-test/r/myisam.result:
  checksums are identical; without the code fix they were all different
mysql-test/t/myisam.test:
  test that same tables give same checksums
sql/sql_table.cc:
  Type GEOMETRY is implemented on top of type BLOB, so, just like for BLOB, its 'field' contains pointers
  which it does not make sense to include in the checksum; it rather has to be converted to a string and
  then we can compute the checksum.

1 files changed, 15 insertions, 0 deletions

diff --git a/mysql-test/r/myisam.result b/mysql-test/r/myisam.result
index 3a0cbd3c32d..4fbe134156f 100644
--- a/mysql-test/r/myisam.result
+++ b/mysql-test/r/myisam.result
@@ -2033,4 +2033,19 @@ t1	CREATE TABLE `t1` (
   `c` char(1) DEFAULT NULL
 ) ENGINE=MyISAM DEFAULT CHARSET=latin1 TRANSACTIONAL=1
 drop table t1;
+CREATE TABLE t1 (line LINESTRING NOT NULL) engine=myisam;
+INSERT INTO t1 VALUES (GeomFromText("POINT(0 0)"));
+checksum table t1;
+Table	Checksum
+test.t1	326284887
+CREATE TABLE t2 (line LINESTRING NOT NULL) engine=myisam;
+INSERT INTO t2 VALUES (GeomFromText("POINT(0 0)"));
+checksum table t2;
+Table	Checksum
+test.t2	326284887
+CREATE TABLE t3 select * from t1;
+checksum table t3;
+Table	Checksum
+test.t3	326284887
+drop table t1,t2,t3;
 End of 5.1 tests