Float(n)

Once in a social network was asked how to remove trailing zeros in decimal numbers. This was associated with the preparation of a report, which sums concatenate with the text. In the conditions of the problem was stated limit to two decimal places.

Here’s an example of data and the desired result:

Solutions have been proposed, based on analysis of the line. I also chose the wrong way and proposed the following solution:

SELECT num,
CASE WHEN CAST(num AS INT) = num
           THEN CAST(CAST(num AS INT) AS VARCHAR)
           WHEN CAST(num*10 AS INT) = num*10
           THEN LEFT(CAST(num AS VARCHAR), LEN(CAST(num AS VARCHAR)) - 1)
            WHEN CAST(num*100 AS INT)=num*100
            THEN CAST(num AS VARCHAR)
END fnum
FROM(
SELECT 0.00 as num
UNION ALL SELECT 10.00
UNION ALL SELECT 2.50
UNION ALL SELECT 100
UNION ALL SELECT 11.33
) X

I do not know how much it would still be continued, if one member did not notice that all problems are solved by conversion to the data type FLOAT. Really:

SELECT num, CAST(num AS FLOAT) fnum
FROM(
SELECT 0.00 as num
UNION ALL SELECT 10.00
UNION ALL SELECT 2.50
UNION ALL SELECT 100
UNION ALL SELECT 11.33
) X

However, you need to remember the approximate nature of this type, namely the magnitude of the mantissa in scientific representation of the number.

In accordance with the standard in this type of data specified argument - FLOAT (n), which can take values from 1 to 53. The SQL Server, an argument value in the range 1 – 24, interprets it as 24, which corresponds to the accuracy of 7 digits, and in the range 25 - 53 as 53, which corresponds to the accuracy of 15 digits. The default is 53.

The following example illustrates the above:

SELECT num,
CAST(num AS FLOAT(24)) num_24,
CAST(num AS FLOAT(53)) num_53
FROM(
SELECT 1234567.80 AS num
UNION ALL SELECT  12345678.90
UNION ALL SELECT 123456789012345.60
UNION ALL SELECT 1234567890123456.70
) x

MySQL (version 5.0)

Does not support the conversion to type FLOAT.

PostgreSQL (version 8.3.6)

Almost similar behavior, except that for the parameter in the range 1 - 24 precision is 6 digits. Accordingly, recent results will look like this: