queries with the same plan have different plan_digest in information_schema.statement_summary

[qw4990]

Bug Report

Please answer these questions before submitting your issue. Thanks!

1. Minimal reproduce step (Required)

create table t (a int);
select * from t where a in (1, 2);
select * from t where a in (1, 2, 3);
use information_schema;

mysql> select plan_digest, digest_text, QUERY_SAMPLE_TEXT from STATEMENTS_SUMMARY where digest_text="select * from `t` where `a` in ( ... )";
+------------------------------------------------------------------+----------------------------------------+--------------------------------------+
| plan_digest                                                      | digest_text                            | QUERY_SAMPLE_TEXT                    |
+------------------------------------------------------------------+----------------------------------------+--------------------------------------+
| ddbf412c92228a757994b14210801d86f61eb9b510e6bc36dca75d5d4151ad02 | select * from `t` where `a` in ( ... ) | select * from t where a in (1, 2, 3) |
| 01308613e49648a4431351439f0596ce1236bc671f9ced66c04bc19b2b8b2c38 | select * from `t` where `a` in ( ... ) | select * from t where a in (1, 2)    |
+------------------------------------------------------------------+----------------------------------------+--------------------------------------+
2 rows in set (0.00 sec)

2. What did you expect to see? (Required)

These two queries above actually have the same plan, so they should have the same plan_digest as well.

3. What did you see instead (Required)

They have different plan_digests.

4. What is your TiDB version? (Required)

mysql> select version();
+-----------------------------------------------+
| version()                                     |
+-----------------------------------------------+
| 5.7.25-TiDB-v6.1.0-alpha-78-g8f8dfafc10-dirty |
+-----------------------------------------------+
1 row in set (0.00 sec)

5. Root Cause Analysis

It is caused by selection operators with different in-list lengths are normalized to different strings.

// ExplainNormalizedInfo implements Plan interface.
func (p *PhysicalSelection) ExplainNormalizedInfo() string {
	return string(expression.SortedExplainNormalizedExpressionList(p.Conditions))
}

[qw4990]

PTAL @djshow832

[likzn]

Is the case meet expection? I found these two sql have different plan.So the plan_digest is different?

select * from t where a in (1, 2);
select * from t where a in (1, 2, 3);

plan->plan_digest
digest_text->digest.
I found digest of the case sqls is the same.