Xlang Type Mapping

Note:

• For type definition, see Type Systems in Spec
• int16_t[n]/vector<T> indicates int16_t[n]/vector<int16_t>
• The cross-language serialization is not stable, do not use it in your production environment.

User Type IDs

When registering user types (struct/ext/enum/union), the internal type ID is written as the 8-bit kind, and the user type ID is written separately as an unsigned varint32. There is no bit shift/packing, and user_type_id can be in the range 0~0xFFFFFFFE.

Examples:

User ID	Type	Internal ID	Encoded User ID	Decimal
0	STRUCT	27	0	0
0	ENUM	25	0	0
1	STRUCT	27	1	1
1	COMPATIBLE_STRUCT	28	1	1
2	NAMED_STRUCT	29	2	2

When reading type IDs:

• Read internal type ID from the type ID field.
• If the internal type is a user-registered kind, read user_type_id as varuint32.

Type Mapping

Fory Type	Fory Type ID	Java	Python	Javascript	C++	Golang	Rust
bool	1	bool/Boolean	bool	Boolean	bool	bool	bool
int8	2	byte/Byte	int/pyfory.int8	Type.int8()	int8_t	int8	i8
int16	3	short/Short	int/pyfory.int16	Type.int16()	int16_t	int16	i16
int32	4	int/Integer	int/pyfory.fixed_int32	Type.int32()	int32_t	int32	i32
varint32	5	int/Integer	int/pyfory.int32	Type.varint32()	int32_t	int32	i32
int64	6	long/Long	int/pyfory.fixed_int64	Type.int64()	int64_t	int64	i64
varint64	7	long/Long	int/pyfory.int64	Type.varint64()	int64_t	int64	i64
tagged_int64	8	long/Long	int/pyfory.tagged_int64	Type.tagged_int64()	int64_t	int64	i64
uint8	9	short/Short	int/pyfory.uint8	Type.uint8()	uint8_t	uint8	u8
uint16	10	int/Integer	int/pyfory.uint16	Type.uint16()	uint16_t	uint16	u16
uint32	11	long/Long	int/pyfory.fixed_uint32	Type.uint32()	uint32_t	uint32	u32
var_uint32	12	long/Long	int/pyfory.uint32	Type.varUInt32()	uint32_t	uint32	u32
uint64	13	long/Long	int/pyfory.fixed_uint64	Type.uint64()	uint64_t	uint64	u64
var_uint64	14	long/Long	int/pyfory.uint64	Type.varUInt64()	uint64_t	uint64	u64
tagged_uint64	15	long/Long	int/pyfory.tagged_uint64	Type.taggedUInt64()	uint64_t	uint64	u64
float8	16	/	/	/	/	/	/
float16	17	float/Float	float/pyfory.float16	Type.float16()	fory::float16_t	fory.float16	fory::f16
bfloat16	18	/	/	/	/	/	/
float32	19	float/Float	float/pyfory.float32	Type.float32()	float	float32	f32
float64	20	double/Double	float/pyfory.float64	Type.float64()	double	float64	f64
string	21	String	str	String	string	string	String/str
list	22	List/Collection	list/tuple	array	vector	slice	Vec
set	23	Set	set	/	set	fory.Set	Set
map	24	Map	dict	Map	unordered_map	map	HashMap
enum	25	Enum subclasses	enum subclasses	/	enum	/	enum
named_enum	26	Enum subclasses	enum subclasses	/	enum	/	enum
struct	27	pojo/record	data class	object	struct/class	struct	struct
compatible_struct	28	pojo/record	data class	object	struct/class	struct	struct
named_struct	29	pojo/record	data class	object	struct/class	struct	struct
named_compatible_struct	30	pojo/record	data class	object	struct/class	struct	struct
ext	31	pojo/record	data class	object	struct/class	struct	struct
named_ext	32	pojo/record	data class	object	struct/class	struct	struct
union	33	Union	typing.Union	/	std::variant<Ts...>	/	tagged union enum
none	36	null	None	null	std::monostate	nil	()
duration	37	Duration	timedelta	Number	duration	Duration	Duration
timestamp	38	Instant	datetime	Number	std::chrono::nanoseconds	Time	DateTime
date	39	Date	datetime	Number	fory::serialization::Date	Time	DateTime
decimal	40	BigDecimal	Decimal	bigint	/	/	/
binary	41	byte[]	bytes	/	uint8_t[n]/vector<T>	[n]uint8/[]T	Vec<uint8_t>
array	42	array	np.ndarray	/	/	array/slice	Vec
bool_array	43	bool[]	ndarray(np.bool_)	/	bool[n]	[n]bool/[]T	Vec<bool>
int8_array	44	byte[]	ndarray(int8)	/	int8_t[n]/vector<T>	[n]int8/[]T	Vec<i8>
int16_array	45	short[]	ndarray(int16)	/	int16_t[n]/vector<T>	[n]int16/[]T	Vec<i16>
int32_array	46	int[]	ndarray(int32)	/	int32_t[n]/vector<T>	[n]int32/[]T	Vec<i32>
int64_array	47	long[]	ndarray(int64)	/	int64_t[n]/vector<T>	[n]int64/[]T	Vec<i64>
uint8_array	48	short[]	ndarray(uint8)	/	uint8_t[n]/vector<T>	[n]uint8/[]T	Vec<u8>
uint16_array	49	int[]	ndarray(uint16)	/	uint16_t[n]/vector<T>	[n]uint16/[]T	Vec<u16>
uint32_array	50	long[]	ndarray(uint32)	/	uint32_t[n]/vector<T>	[n]uint32/[]T	Vec<u32>
uint64_array	51	long[]	ndarray(uint64)	/	uint64_t[n]/vector<T>	[n]uint64/[]T	Vec<u64>
float8_array	52	/	/	/	/	/	/
float16_array	53	float[]	ndarray(float16)	/	fory::float16_t[n]/vector<T>	[n]float16/[]T	Vec<fory::f16>
bfloat16_array	54	/	/	/	/	/	/
float32_array	55	float[]	ndarray(float32)	/	float[n]/vector<T>	[n]float32/[]T	Vec<f32>
float64_array	56	double[]	ndarray(float64)	/	double[n]/vector<T>	[n]float64/[]T	Vec<f64>

Type info(not implemented currently)

Due to differences between type systems of languages, those types can't be mapped one-to-one between languages.

If the user notices that one type on a language corresponds to multiple types in Fory type systems, for example, long in java has type int64/varint64/h64, it means the language lacks some types, and the user must provide extra type info when using Fory.

Type annotation

If the type is a field of another class, users can provide meta hints for fields of a type, or for the whole type. Such information can be provided in other languages too:

• java: use annotation.
• cpp: use macro and template.
• golang: use struct tag.
• python: use typehint.
• rust: use macro.

Here is en example:

• Java:

  class Foo {
    @Int32Type(varint = true)
    int f1;
    List<@Int32Type(varint = true) Integer> f2;
  }

• Python:

  class Foo:
      f1: pyfory.varint32
      f2: List[pyfory.varint32]