sint32sint64

Sint32 and sint64 are signed integer types used in Protocol Buffers (protobuf) with a specific encoding designed to optimize space for values with small magnitude. The pair—often discussed together as sint32 and sint64—serves as the signed counterparts to the unsigned and fixed-width integer types in protobuf, but they employ a different encoding approach to reduce the size of negative numbers that are near zero.

Both types use zigzag encoding before applying varint encoding. Zigzag encoding converts the signed integer to an unsigned value in such a way that small magnitude numbers (both positive and negative) receive small unsigned representations. For 32-bit values, the mapping is (n << 1) ^ (n >> 31); for 64-bit values, it is (n << 1) ^ (n >> 63). After zigzag transformation, protobuf uses varint encoding, which encodes the resulting unsigned value in one to five bytes for sint32 and one to ten bytes for sint64, depending on the magnitude.

The practical effect is that sint32 and sint64 typically compress better than the standard signed int32 and int64 representations when the data set contains many small, especially negative, numbers. This makes sint types advantageous for certain data models, such as time durations, deltas, or counts that cluster near zero.

Ranges for the value domain align with their bit width: sint32 covers -2,147,483,648 to 2,147,483,647, and sint64 covers -9,223,372,036,854,775,808 to 9,223,372,036,854,775,807. In protobuf schemas, sint32 and sint64 are chosen when the expected value distribution benefits from zigzag-encoded varints. They are part of proto2 and proto3 syntax and can be used alongside other integer types like int32, int64, sfixed32, and sfixed64 as appropriate for the data model.