VARCHAR

A variable-length ASCII text string. Use VARCHAR for columns that contain only ASCII characters. For text with accented characters, emoji, or any non-ASCII content, use NVARCHAR instead.

Aliases: STRING, TEXT

Example

SELECT 'hello world';

Casting

From	Example	Notes
from INTEGER	`42::VARCHAR`	Decimal string representation
from FLOAT	`3.14::VARCHAR`	Decimal notation
from BOOLEAN	`TRUE::VARCHAR`	'true' or 'false'
from DATE	`date_col::VARCHAR`	'YYYY-MM-DD'
from TIMESTAMP	`ts_col::VARCHAR`	ISO 8601 string representation
from VARBINARY	`bin_col::VARCHAR`	Reinterprets the raw bytes as ASCII text. UTF-8 is NOT validated, so bytes that are not ASCII produce an undefined VARCHAR that a client may be unable to decode — cast to NVARCHAR instead, which validates and fails on invalid sequences

Comparisons

Can be compared (using =, <, >, etc.) with: VARCHAR, NVARCHAR, VARBINARY.

Notes

Supports LIKE (case-sensitive), ILIKE (case-insensitive), and RLIKE (regular expression) pattern matching. Every VARCHAR operation works on BYTES: LENGTH counts bytes, SUBSTRING/LEFT/RIGHT take byte offsets, and REVERSE reverses bytes. On ASCII content — the only content VARCHAR defines — bytes and characters are the same thing. Use NVARCHAR for anything else; its equivalents count code points.

Limitations

Non-ASCII bytes stored in a VARCHAR column produce undefined behaviour — use NVARCHAR for Unicode. A byte-wise operation will happily split a multi-byte character: REVERSE over UTF-8 held in a VARCHAR returns an invalid sequence, and that is the type's contract, not a fault in the function.
String values (VARCHAR/NVARCHAR/VARBINARY alike) are length-capped at just under 4 GiB per value (length is stored as uint32).