Numeric Type Literal
Use numeric Literals to specify fixed-point and floating-point numbers.
Integer Literal
Integers are represented as a sequence of digits. They may have a sign. For example: +1, -2, 345.
Doris determines the type used to store integer literals based on the input value. The range mapping is shown in the table below:
| Value Range | Type |
|---|---|
| -2^7 to 2^7 - 1 | TINYINT |
| -2^15 to 2^15 - 1 | SMALLINT |
| -2^31 to 2^31 - 1 | INT |
| -2^63 to 2^63 - 1 | BIGINT |
| -2^127 to 2^127 - 1 | LARGEINT |
Fixed and Floating Point Numeric Literals
Fixed and floating point numeric literals can have an integer part, a fractional part, or both. They may have a sign. For example: 1, .2, 3.4, -5, -6.78, +9.10.
They can also be represented in scientific notation, including the significand and exponent. Either part or both can have a sign. For example: 1.2E3, 1.2E-3, -1.2E3, -1.2E-3.
Numbers represented in this way are preferentially parsed as fixed-point numbers. The range of supported fixed-point numbers is controlled by the variable enable_decimal256. When enable_decimal256 is TRUE, the maximum precision is 76. When enable_decimal256 is FALSE, the maximum precision is 38.
When the precision required by a number exceeds the maximum value representable by fixed-point numbers, it is parsed as a floating-point number, with the type DOUBLE.