Data types

Data manipulation language (DML) operations on tables

UNION, INTERSECT, and EXCEPT queries

CASE expressions

Evaluation of predicates, such as LIKE and IN

Evaluation of SQL functions that do comparisons or extractions of data

Comparisons with mathematical operators

In general, data types that fall into the same type category (such as different numeric data types) are compatible and can be implicitly converted.

For example, with implicit conversion you can insert a decimal value into an integer column. The decimal is rounded to produce a whole number. Or you can extract a numeric value, such as 2008, from a date and insert that value into an integer column.

Numeric data types enforce overflow conditions that occur when you attempt to insert out-of-range values. For example, a decimal value with a precision of 5 does not fit into a decimal column that was defined with a precision of 4. An integer or the whole part of a decimal is never truncated; however, the fractional part of a decimal can be rounded up or down, as appropriate. However, results of explicit casts of values selected from tables are not rounded.

Different types of character strings are compatible; VARCHAR column strings containing single-byte data and CHAR column strings are comparable and implicitly convertible. VARCHAR strings that contain multibyte data are not comparable. Also, you can convert a character string to a date, time, timestamp, or numeric value if the string is an appropriate literal value; any leading or trailing spaces are ignored. Conversely, you can convert a date, time, timestamp, or numeric value to a fixed-length or variable-length character string.

If you compare DECIMAL values with character strings, Amazon Redshift attempts to convert the character string to a DECIMAL value. When comparing all other numeric values with character strings, the numeric values are converted to character strings. To enforce the opposite conversion (for example, converting character strings to integers, or converting DECIMAL values to character strings), use an explicit function, such as CAST.

To convert 64-bit DECIMAL or NUMERIC values to a higher precision, you must use an explicit conversion function such as the CAST or CONVERT functions.

When converting DATE or TIMESTAMP to TIMESTAMPTZ, or converting TIME to TIMETZ, the time zone is set to the current session time zone. The session time zone is UTC by default. For more information about setting the session time zone, see timezone.

Similarly, TIMESTAMPTZ is converted to DATE, TIME, or TIMESTAMP based on the current session time zone. The session time zone is UTC by default. After the conversion, time zone information is dropped.

Character strings that represent a timestamp with time zone specified are converted to TIMESTAMPTZ using the current session time zone, which is UTC by default. Likewise, character strings that represent a time with time zone specified are converted to TIMETZ using the current session time zone, which is UTC by default.

Implicit conversions in assignments, such as setting values in INSERT or UPDATE commands.

Implicit conversions in expressions, such as performing comparisons in the WHERE clause.