1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
|
-- source include/have_ndb.inc
-- source include/not_embedded.inc
--disable_warnings
drop table if exists t1, test1, test2;
--enable_warnings
#
# Simple test to show use of ordered indexes
#
CREATE TABLE t1 (
a int unsigned NOT NULL PRIMARY KEY,
b int unsigned not null,
c int unsigned,
KEY(b)
) engine=ndbcluster;
insert t1 values(1, 2, 3), (2,3, 5), (3, 4, 6), (4, 5, 8), (5,6, 2), (6,7, 2);
select * from t1 order by b;
select * from t1 where b >= 4 order by b;
select * from t1 where b = 4 order by b;
select * from t1 where b > 4 order by b;
select * from t1 where b < 4 order by b;
select * from t1 where b <= 4 order by b;
# Test of reset_bounds
select tt1.* from t1 as tt1, t1 as tt2 use index(b) where tt1.b = tt2.b order by tt1.b;
select a, b, c from t1 where a!=2 and c=6;
select a, b, c from t1 where a!=2 order by a;
#
# Here we should add some "explain select" to verify that the ordered index is
# used for these queries.
#
#
# Update using ordered index scan
#
update t1 set c = 3 where b = 3;
select * from t1 order by a;
update t1 set c = 10 where b >= 6;
select * from t1 order by a;
update t1 set c = 11 where b < 5;
select * from t1 order by a;
update t1 set c = 12 where b > 0;
select * from t1 order by a;
update t1 set c = 13 where b <= 3;
select * from t1 order by a;
update t1 set b = b + 1 where b > 4 and b < 7;
select * from t1 order by a;
-- Update primary key
update t1 set a = a + 10 where b > 1 and b < 7;
select * from t1 order by a;
#
# Delete using ordered index scan
#
drop table t1;
CREATE TABLE t1 (
a int unsigned NOT NULL PRIMARY KEY,
b int unsigned not null,
c int unsigned,
KEY(b)
) engine=ndbcluster;
insert t1 values(1, 2, 13), (2,3, 13), (3, 4, 12), (4, 5, 12), (5,6, 12), (6,7, 12);
delete from t1 where b = 3;
select * from t1 order by a;
delete from t1 where b >= 6;
select * from t1 order by a;
delete from t1 where b < 4;
select * from t1 order by a;
delete from t1 where b > 5;
select * from t1 order by a;
delete from t1 where b <= 4;
select * from t1 order by a;
drop table t1;
#
#multi part key
#
CREATE TABLE t1 (
a int unsigned NOT NULL PRIMARY KEY,
b int unsigned not null,
c int unsigned not null
) engine = ndb;
create index a1 on t1 (b, c);
insert into t1 values (1, 2, 13);
insert into t1 values (2,3, 13);
insert into t1 values (3, 4, 12);
insert into t1 values (4, 5, 12);
insert into t1 values (5,6, 12);
insert into t1 values (6,7, 12);
insert into t1 values (7, 2, 1);
insert into t1 values (8,3, 6);
insert into t1 values (9, 4, 12);
insert into t1 values (14, 5, 4);
insert into t1 values (15,5,5);
insert into t1 values (16,5, 6);
insert into t1 values (17,4,4);
insert into t1 values (18,1, 7);
select * from t1 order by a;
select * from t1 where b<=5 order by a;
select * from t1 where b<=5 and c=0;
insert into t1 values (19,4, 0);
select * from t1 where b<=5 and c=0;
select * from t1 where b=4 and c<=5 order by a;
select * from t1 where b<=4 and c<=5 order by a;
select * from t1 where b<=5 and c=0 or b<=5 and c=2;
select count(*) from t1 where b = 0;
select count(*) from t1 where b = 1;
drop table t1;
#
# Indexing NULL values
#
CREATE TABLE t1 (
a int unsigned NOT NULL PRIMARY KEY,
b int unsigned,
c int unsigned,
KEY bc(b,c)
) engine = ndb;
insert into t1 values(1,1,1),(2,NULL,2),(3,NULL,NULL),(4,4,NULL);
select * from t1 use index (bc) where b IS NULL order by a;
select * from t1 use index (bc)order by a;
select * from t1 use index (bc) order by a;
select * from t1 use index (PRIMARY) where b IS NULL order by a;
select * from t1 use index (bc) where b IS NULL order by a;
select * from t1 use index (bc) where b IS NULL and c IS NULL order by a;
select * from t1 use index (bc) where b IS NULL and c = 2 order by a;
select * from t1 use index (bc) where b < 4 order by a;
select * from t1 use index (bc) where b IS NOT NULL order by a;
drop table t1;
#
# Order by again, including descending.
#
create table t1 (
a int unsigned primary key,
b int unsigned,
c char(10),
key bc (b, c)
) engine=ndb;
insert into t1 values(1,1,'a'),(2,2,'b'),(3,3,'c'),(4,4,'d'),(5,5,'e');
insert into t1 select a*7,10*b,'f' from t1;
insert into t1 select a*13,10*b,'g' from t1;
insert into t1 select a*17,10*b,'h' from t1;
insert into t1 select a*19,10*b,'i' from t1;
insert into t1 select a*23,10*b,'j' from t1;
insert into t1 select a*29,10*b,'k' from t1;
#
select b, c from t1 where b <= 10 and c <'f' order by b, c;
select b, c from t1 where b <= 10 and c <'f' order by b desc, c desc;
#
select b, c from t1 where b=4000 and c<'k' order by b, c;
select b, c from t1 where b=4000 and c<'k' order by b desc, c desc;
select b, c from t1 where 1000<=b and b<=100000 and c<'j' order by b, c;
select b, c from t1 where 1000<=b and b<=100000 and c<'j' order by b desc, c desc;
#
select min(b), max(b) from t1;
#
drop table t1;
#
# Bug #6435
CREATE TABLE test1 (
SubscrID int(11) NOT NULL auto_increment,
UsrID int(11) NOT NULL default '0',
PRIMARY KEY (SubscrID),
KEY idx_usrid (UsrID)
) ENGINE=ndbcluster DEFAULT CHARSET=latin1;
INSERT INTO test1 VALUES (2,224),(3,224),(1,224);
CREATE TABLE test2 (
SbclID int(11) NOT NULL auto_increment,
SbcrID int(11) NOT NULL default '0',
PRIMARY KEY (SbclID),
KEY idx_sbcrid (SbcrID)
) ENGINE=ndbcluster DEFAULT CHARSET=latin1;
INSERT INTO test2 VALUES (3,2),(1,1),(2,1),(4,2);
select * from test1 order by 1;
select * from test2 order by 1;
SELECT s.SubscrID,l.SbclID FROM test1 s left JOIN test2 l ON
l.SbcrID=s.SubscrID WHERE s.UsrID=224 order by 1, 2;
drop table test1;
drop table test2;
# bug#7424 + bug#7725
create table t1 (
pk int primary key,
dt datetime not null,
da date not null,
ye year not null,
ti time not null,
ts timestamp not null,
index(dt),
index(da),
index(ye),
index(ti),
index(ts)
) engine=ndb;
insert into t1 (pk,dt,da,ye,ti,ts) values
(1, '1901-05-05 23:00:59', '1901-05-05', '1901', '23:00:59', '2001-01-01 23:00:59'),
(2, '1912-09-05 13:00:59', '1912-09-05', '1912', '13:00:59', '2001-01-01 13:00:59'),
(3, '1945-12-31 00:00:00', '1945-12-31', '1945', '00:00:00', '2001-01-01 00:00:00'),
(4, '1955-12-31 00:00:00', '1955-12-31', '1955', '00:00:00', '2001-01-01 00:00:00'),
(5, '1963-06-06 06:06:06', '1963-06-06', '1963', '06:06:06', '2001-01-01 06:06:06'),
(6, '1993-06-06 06:06:06', '1993-06-06', '1993', '06:06:06', '2001-01-01 06:06:06'),
(7, '2001-01-01 10:11:10', '2001-01-01', '2001', '10:11:10', '2001-01-01 10:11:10'),
(8, '2001-01-01 10:11:11', '2001-01-01', '2001', '10:11:11', '2001-01-01 10:11:11'),
(9, '2005-01-31 23:59:59', '2005-01-31', '2005', '23:59:59', '2001-01-01 23:59:59');
# datetime
select count(*)-9 from t1 use index (dt) where dt > '1900-01-01 00:00:00';
select count(*)-6 from t1 use index (dt) where dt >= '1955-12-31 00:00:00';
select count(*)-5 from t1 use index (dt) where dt > '1955-12-31 00:00:00';
select count(*)-5 from t1 use index (dt) where dt < '1970-03-03 22:22:22';
select count(*)-7 from t1 use index (dt) where dt < '2001-01-01 10:11:11';
select count(*)-8 from t1 use index (dt) where dt <= '2001-01-01 10:11:11';
select count(*)-9 from t1 use index (dt) where dt <= '2055-01-01 00:00:00';
# date
select count(*)-9 from t1 use index (da) where da > '1900-01-01';
select count(*)-6 from t1 use index (da) where da >= '1955-12-31';
select count(*)-5 from t1 use index (da) where da > '1955-12-31';
select count(*)-5 from t1 use index (da) where da < '1970-03-03';
select count(*)-6 from t1 use index (da) where da < '2001-01-01';
select count(*)-8 from t1 use index (da) where da <= '2001-01-02';
select count(*)-9 from t1 use index (da) where da <= '2055-01-01';
# year
select count(*)-9 from t1 use index (ye) where ye > '1900';
select count(*)-6 from t1 use index (ye) where ye >= '1955';
select count(*)-5 from t1 use index (ye) where ye > '1955';
select count(*)-5 from t1 use index (ye) where ye < '1970';
select count(*)-6 from t1 use index (ye) where ye < '2001';
select count(*)-8 from t1 use index (ye) where ye <= '2001';
select count(*)-9 from t1 use index (ye) where ye <= '2055';
# time
select count(*)-9 from t1 use index (ti) where ti >= '00:00:00';
select count(*)-7 from t1 use index (ti) where ti > '00:00:00';
select count(*)-7 from t1 use index (ti) where ti > '05:05:05';
select count(*)-5 from t1 use index (ti) where ti > '06:06:06';
select count(*)-5 from t1 use index (ti) where ti < '10:11:11';
select count(*)-6 from t1 use index (ti) where ti <= '10:11:11';
select count(*)-8 from t1 use index (ti) where ti < '23:59:59';
select count(*)-9 from t1 use index (ti) where ti <= '23:59:59';
# timestamp
select count(*)-9 from t1 use index (ts) where ts >= '2001-01-01 00:00:00';
select count(*)-7 from t1 use index (ts) where ts > '2001-01-01 00:00:00';
select count(*)-7 from t1 use index (ts) where ts > '2001-01-01 05:05:05';
select count(*)-5 from t1 use index (ts) where ts > '2001-01-01 06:06:06';
select count(*)-5 from t1 use index (ts) where ts < '2001-01-01 10:11:11';
select count(*)-6 from t1 use index (ts) where ts <= '2001-01-01 10:11:11';
select count(*)-8 from t1 use index (ts) where ts < '2001-01-01 23:59:59';
select count(*)-9 from t1 use index (ts) where ts <= '2001-01-01 23:59:59';
drop table t1;
# decimal (not the new 5.0 thing)
create table t1 (
a int primary key,
s decimal(12),
t decimal(12, 5),
u decimal(12) unsigned,
v decimal(12, 5) unsigned,
key (s),
key (t),
key (u),
key (v)
) engine=ndb;
#
insert into t1 values
( 0, -000000000007, -0000061.00003, 000000000061, 0000965.00042),
( 1, -000000000007, -0000061.00042, 000000000061, 0000965.00003),
( 2, -071006035767, 4210253.00024, 000000000001, 0000001.84488),
( 3, 000000007115, 0000000.77607, 000077350625, 0000018.00013),
( 4, -000000068391, -0346486.00000, 000000005071, 0005334.00002),
( 5, -521579890459, -1936874.00001, 000000000154, 0000003.00018),
( 6, -521579890459, -1936874.00018, 000000000154, 0000003.00001),
( 7, 000000000333, 0000051.39140, 000000907958, 0788643.08374),
( 8, 000042731229, 0000009.00000, 000000000009, 6428667.00000),
( 9, -000008159769, 0000918.00004, 000096951421, 7607730.00008);
#
select count(*)- 5 from t1 use index (s) where s < -000000000007;
select count(*)- 7 from t1 use index (s) where s <= -000000000007;
select count(*)- 2 from t1 use index (s) where s = -000000000007;
select count(*)- 5 from t1 use index (s) where s >= -000000000007;
select count(*)- 3 from t1 use index (s) where s > -000000000007;
#
select count(*)- 4 from t1 use index (t) where t < -0000061.00003;
select count(*)- 5 from t1 use index (t) where t <= -0000061.00003;
select count(*)- 1 from t1 use index (t) where t = -0000061.00003;
select count(*)- 6 from t1 use index (t) where t >= -0000061.00003;
select count(*)- 5 from t1 use index (t) where t > -0000061.00003;
#
select count(*)- 2 from t1 use index (u) where u < 000000000061;
select count(*)- 4 from t1 use index (u) where u <= 000000000061;
select count(*)- 2 from t1 use index (u) where u = 000000000061;
select count(*)- 8 from t1 use index (u) where u >= 000000000061;
select count(*)- 6 from t1 use index (u) where u > 000000000061;
#
select count(*)- 5 from t1 use index (v) where v < 0000965.00042;
select count(*)- 6 from t1 use index (v) where v <= 0000965.00042;
select count(*)- 1 from t1 use index (v) where v = 0000965.00042;
select count(*)- 5 from t1 use index (v) where v >= 0000965.00042;
select count(*)- 4 from t1 use index (v) where v > 0000965.00042;
drop table t1;
# bug#7798
create table t1(a int primary key, b int not null, index(b));
insert into t1 values (1,1), (2,2);
connect (con1,localhost,,,test);
connect (con2,localhost,,,test);
connection con1;
set autocommit=0;
begin;
select count(*) from t1;
connection con2;
ALTER TABLE t1 ADD COLUMN c int;
connection con1;
select a from t1 where b = 2;
show tables;
drop table t1;
# mysqld 5.0.13 crash, no bug#
create table t1 (a int, c varchar(10),
primary key using hash (a), index(c)) engine=ndb;
insert into t1 (a, c) values (1,'aaa'),(3,'bbb');
select count(*) from t1 where c<'bbb';
# End of 4.1 tests
|