SQL SERVER - how (texto) Comprensión de las obras MIN

Question

Estoy haciendo un poco de investigación y en busca de una explicación sobre cómo el servidor SQL evalúa MIN(Varchar).

He encontrado esta observación en BOL: MIN busca el valor más bajo en el orden de clasificación definido en la base de datos subyacente

Así que si tengo una tabla que tiene una fila con los valores siguientes:

Data

AA
AB
AC

Hacer una SELECT MIN(DATA) volvería AA espalda. Sólo quiero entender el por qué detrás de esto y entender la BOL un poco mejor.

Gracias!

Answer 1

It's determined by the collation (sort order). For most cultures the collation order is the same as the alphabetical order in the English alphabet so:

• 'AA' < 'AB'
• 'AA' < 'AC'
• 'AB' < 'AC'

Therefore 'AA' is the minimum value. For other cultures this may not hold. For example a Danish collation would return 'AB' as the minimum because 'AA' > 'AB'. This is because 'AA' is treated as equivalent to 'Å' which is the last letter in the Danish alphabet.

SELECT MIN(s COLLATE Danish_Norwegian_CI_AS) FROM table1;

min_s
AB

To get an "ordinary" sort order use the Latin1_General_Bin collation:

SELECT MIN(s COLLATE Latin1_General_Bin) FROM table1;

min_s
AA

To reproduce this result you can create this test table:

CREATE TABLE table1 (s varchar(100));
INSERT INTO table1 (s) VALUES ('AA'), ('AB'), ('AC');

Answer 2

No, MIN is used in a SELECT statement that scans more than one line. It takes a column as an argument, and returns the "lowest" value (again, according to the collation sequence) found in that column.

Used without a GROUP BY clause, the result set will have a single row, and the value of MIN will be the lowest value found in that column. Used with a GROUP BY clause, the result set will have one row for each group and the value of MIN will be the lowest value in that column for any row in the group.

Answer 3

min(x), where is a char (string) type -- char(), varchar(), nchar(), nvarchar(), finds the lowest value in the group, based on SQL's string comparison rules:

• if two strings differ in length, the shorter is padded with SP characters (spaces) to the length of the longer.
• comparison proceeds left-to-right, character by character, according to the rule of the collation sequence in use.
• in comparisons, the value NULL compares lower than any non-null values (the ISO/ANSI SQL standard says that it is an implementation choice as to whether NULL collates lower or higher than any non-null value).

So, if you have a table

create table foo
(
  myString varchar(16) not null ,
)

then running the query

select min(myString) from foo

will give you the same result set as if you executed

set rowcount 1

select myString
from foo
order by myString

set rowcount 0

You are basically ordering the set in ascending sequence and selecting the first value. MAX(), or course, gives you the inverse, ordering the set in descending sequence and selecting the first value.