VINTs

VINTs, short for variable-length integers, are a data type used in EBML, the Extensible Binary Meta Language. In EBML and formats based on it, such as the Matroska multimedia container, VINTs are used to encode element IDs and element sizes. The design makes the encoding self-delimiting: the length of a VINT is determined by the pattern of leading bits in the first byte, allowing a parser to know where the value ends without a separate length field.

Encoding rules: The number of leading zero bits in the first byte dictates the total number of bytes in the VINT. A first byte starting with “1” indicates a one-byte VINT; starting with “01” indicates a two-byte VINT; “001” indicates three bytes; and so on up to eight bytes. The value is formed by the data bits of the first byte (after removing the length indicator) combined with the payload bits of subsequent bytes. One-byte VINTs cover values up to 127, two-byte VINTs cover larger ranges, etc.

Usage: In Matroska, IDs are encoded as VINTs, and the sizes of elements are also encoded as VINTs. This enables files to describe a hierarchical structure with a compact representation for small values while accommodating larger values when necessary. VINTs are unsigned and are not designed to carry negative values.

Decoding and compatibility: Most EBML parsers implement a VINT reader that examines the first byte to determine length and then reads the remaining bytes. VINT encoding is specific to EBML and its derivatives; other variable-length integer schemes exist under different names in other formats. See also EBML, Matroska, variable-length integer.