[CASSANDRA-14245] SELECT JSON prints null on empty strings

• Type: Bug
• Status: Resolved
• Priority: Normal
• Resolution: Fixed
• Fix Version/s: 3.11.3, 4.0-alpha1, 4.0
• Component/s: Legacy/CQL

Environment

Cassandra 3.11.2, Ubuntu 16.04 LTS

 

• Severity: Normal
• Since Version: 3.11.0
• Assignee: Francisco Fernandez
• Reporter: Norbert Schultz

Authors

Francisco Fernandez

Reviewers

Benjamin Lerer

• Created: 21/Feb/18 09:11
• Updated: 15/May/20 08:03
• Resolved: 29/Mar/18 06:57

Description

SELECT JSON reports an empty string as null.

 

Example:

cqlsh:unittest> create table test(id INT, name TEXT, PRIMARY KEY(id));
cqlsh:unittest> insert into test (id, name) VALUES (1, 'Foo');
cqlsh:unittest> insert into test (id, name) VALUES (2, '');
cqlsh:unittest> insert into test (id, name) VALUES (3, null);

cqlsh:unittest> select * from test;

id | name
----+------
  1 |  Foo
  2 |     
  3 | null

(3 rows)

cqlsh:unittest> select JSON * from test;

[json]
--------------------------
{"id": 1, "name": "Foo"}
{"id": 2, "name": null}
{"id": 3, "name": null}

(3 rows)

 

This even happens, if the string is part of the Primary Key, which makes the generated string not insertable.

 

cqlsh:unittest> create table test2 (id INT, name TEXT, age INT, PRIMARY KEY(id, name));
cqlsh:unittest> insert into test2 (id, name, age) VALUES (1, '', 42);
cqlsh:unittest> select JSON * from test2;

[json]
------------------------------------
{"id": 1, "name": null, "age": 42}

(1 rows)

cqlsh:unittest> insert into test2 JSON '{"id": 1, "name": null, "age": 42}';
InvalidRequest: Error from server: code=2200 [Invalid query] message="Invalid null value in condition for column name"

 

On an older version of Cassandra (3.0.8) does not have this problem.