Question

I have a rank table and its schema is defined as rank(id, key,value) where key is the primary key

sqlite> .schema rank
CREATE TABLE 'rank' ( ID VARCHAR, KEY VARCHAR, VALUE VARCHAR NOT NULL );
CREATE INDEX 'rank_id___djklaf3451jlZZZRFfa___' ON 'rank' ( ID );

sqlite> select * from rank;
-----------------------------------------------------
Tymb-W64uwvM|Tymc8LPQnxBg|5
TymdPRdFpBcE|TymdSsaIFhuI|2
TymdPRdFpBcE|TymdjkGgExcE|3
TymeVf6N1RH4|TymeZGxydCJ8|3
TymeVf6N1RH4|Tymeecz1ORW8|20

sqlite> select id,min(value) from rank group by id;
------------------------------------------------------
Tymb-W64uwvM|5
TymdPRdFpBcE|2
TymeVf6N1RH4|20

sqlite> select id,max(value) from rank group by id;
------------------------------------------------------
Tymb-W64uwvM|5
TymdPRdFpBcE|3
TymeVf6N1RH4|3

As you can see the 3rd result for both the min and max functions are incorrect. The sqlite version is 3.6.23.

Any advice?

Answer 1

It’s because value is a character type (specifically varchar) rather than a numeric type.

If you’re sorting character data, it comes out as:

2
20
3
3
5

which is what you’d see if you executed:

select value from rank order by value asc

If you want it to sort numerically, the column should be defined as a numeric type, such as integer.