版本：0.10

Fury Xlang Serialization Format

Cross-language Serialization Specification

Format Version History:

Version 0.1 - serialization spec formalized

Fury xlang serialization is an automatic object serialization framework that supports reference and polymorphism. Fury will convert an object from/to fury xlang serialization binary format. Fury has two core concepts for xlang serialization:

Fury xlang binary format
Framework implemented in different languages to convert object to/from Fury xlang binary format

The serialization format is a dynamic binary format. The dynamics and reference/polymorphism support make Fury flexible, much more easy to use, but also introduce more complexities compared to static serialization frameworks. So the format will be more complex.

Type Systems

Data Types

bool: a boolean value (true or false).
int8: a 8-bit signed integer.
int16: a 16-bit signed integer.
int32: a 32-bit signed integer.
var_int32: a 32-bit signed integer which use fury var_int32 encoding.
int64: a 64-bit signed integer.
var_int64: a 64-bit signed integer which use fury PVL encoding.
sli_int64: a 64-bit signed integer which use fury SLI encoding.
float16: a 16-bit floating point number.
float32: a 32-bit floating point number.
float64: a 64-bit floating point number including NaN and Infinity.
string: a text string encoded using Latin1/UTF16/UTF-8 encoding.
enum: a data type consisting of a set of named values. Rust enum with non-predefined field values are not supported as an enum.
named_enum: an enum whose value will be serialized as the registered name.
struct: a morphic(final) type serialized by Fury Struct serializer. i.e. it doesn't have subclasses. Suppose we're deserializing List<SomeClass>, we can save dynamic serializer dispatch since SomeClass is morphic(final).
compatible_struct: a morphic(final) type serialized by Fury compatible Struct serializer.
named_struct: a struct whose type mapping will be encoded as a name.
named_compatible_struct: a compatible_struct whose type mapping will be encoded as a name.
ext: a type which will be serialized by a customized serializer.
named_ext: an ext type whose type mapping will be encoded as a name.
list: a sequence of objects.
set: an unordered set of unique elements.
map: a map of key-value pairs. Mutable types such as list/map/set/array/tensor/arrow are not allowed as key of map.
duration: an absolute length of time, independent of any calendar/timezone, as a count of nanoseconds.
timestamp: a point in time, independent of any calendar/timezone, as a count of nanoseconds. The count is relative to an epoch at UTC midnight on January 1, 1970.
local_date: a naive date without timezone. The count is days relative to an epoch at UTC midnight on Jan 1, 1970.
decimal: exact decimal value represented as an integer value in two's complement.
binary: an variable-length array of bytes.
array: only allow 1d numeric components. Other arrays will be taken as List. The implementation should support the interoperability between array and list.
- bool_array: one dimensional int16 array.
- int8_array: one dimensional int8 array.
- int16_array: one dimensional int16 array.
- int32_array: one dimensional int32 array.
- int64_array: one dimensional int64 array.
- float16_array: one dimensional half_float_16 array.
- float32_array: one dimensional float32 array.
- float64_array: one dimensional float64 array.
tensor: multidimensional array which every sub-array have same size and type.
arrow record batch: an arrow record batch object.
arrow table: an arrow table object.

Note:

Unsigned int/long are not added here, since not every language support those types.

Polymorphisms

For polymorphism, if one non-final class is registered, and only one subclass is registered, then we can take all elements in List/Map have same type, thus reduce runtime check cost.

Collection/Array polymorphism are not fully supported, since some languages such as golang have only one collection type. If users want to get exactly the type he passed, he must pass that type when deserializing or annotate that type to the field of struct.

Type disambiguation

Due to differences between type systems of languages, those types can't be mapped one-to-one between languages. When deserializing, Fury use the target data structure type and the data type in the data jointly to determine how to deserialize and populate the target data structure. For example:

class Foo {
  int[] intArray;
  Object[] objects;
  List<Object> objectList;
}

class Foo2 {
  int[] intArray;
  List<Object> objects;
  List<Object> objectList;
}

intArray has an int32_array type. But both objects and objectList fields in the serialize data have list data type. When deserializing, the implementation will create an Object array for objects, but create a ArrayList for objectList to populate its elements. And the serialized data of Foo can be deserialized into Foo2 too.

Users can also provide meta hints for fields of a type, or the type whole. Here is an example in java which use annotation to provide such information.

@FuryObject(fieldsNullable = false, trackingRef = false)
class Foo {
  @FuryField(trackingRef = false)
  int[] intArray;
  @FuryField(polymorphic = true)
  Object object;
  @FuryField(tagId = 1, nullable = true)
  List<Object> objectList;
}

Such information can be provided in other languages too:

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

Type ID

All internal data types are expressed using an ID in range 0~64. Users can use 0~4096 for representing their types.

Type mapping

See Type mapping

Spec overview

Here is the overall format:

| fury header | object ref meta | object type meta | object value data |

The data are serialized using little endian byte order overall. If bytes swap is costly for some object, Fury will write the byte order for that object into the data instead of converting it to little endian.

Fury header

Fury header consists starts one byte:

|    2 bytes   |     4 bits    | 1 bit | 1 bit | 1 bit  | 1 bit |   1 byte   |          optional 4 bytes          |
+--------------+---------------+-------+-------+--------+-------+------------+------------------------------------+
| magic number | reserved bits |  oob  | xlang | endian | null  |  language  | unsigned int for meta start offset |

magic number: used to identify fury serialization protocol, current version use 0x62d4.
null flag: 1 when object is null, 0 otherwise. If an object is null, other bits won't be set.
endian flag: 1 when data is encoded by little endian, 0 for big endian.
xlang flag: 1 when serialization uses xlang format, 0 when serialization uses Fury java format.
oob flag: 1 when passed BufferCallback is not null, 0 otherwise.
language: the language when serializing objects, such as JAVA, PYTHON, GO, etc. Fury can use this flag to determine whether spend more time on serialization to make the deserialization faster for dynamic languages.

If meta share mode is enabled, an uncompressed unsigned int is appended to indicate the start offset of metadata.

Reference Meta

Reference tracking handles whether the object is null, and whether to track reference for the object by writing corresponding flags and maintaining internal state.

Reference flags:

Flag	Byte Value	Description
NULL FLAG	`-3`	This flag indicates the object is a null value. We don't use another byte to indicate REF, so that we can save one byte.
REF FLAG	`-2`	This flag indicates the object is already serialized previously, and fury will write a ref id with unsigned varint format instead of serialize it again
NOT_NULL VALUE FLAG	`-1`	This flag indicates the object is a non-null value and fury doesn't track ref for this type of object.
REF VALUE FLAG	`0`	This flag indicates the object is referencable and the first time to serialize.

When reference tracking is disabled globally or for specific types, or for certain types within a particular context(e.g., a field of a type), only the NULL and NOT_NULL VALUE flags will be used for reference meta.

For languages which doesn't support reference such as rust, reference tracking must be disabled for correct deserialization by fury rust implementation.

For languages whose object values are not null by default:

In rust, Fury takes Option:None as a null value
In c++, Fury takes std::nullopt as a null value
In golang, Fury takes null interface/pointer as a null value

If one want to deserialize in languages like Java/Python/JavaScript, he should mark the type with all fields not-null by default, or using schema-evolution mode to carry the not-null fields info in the data.

Type Meta

For every type to be serialized, it have a type id to indicate its type.

basic types: the type id
enum:
- Type.ENUM + registered id
- Type.NAMED_ENUM + registered namespace+typename
list: Type.List
set: Type.SET
map: Type.MAP
ext:
- Type.EXT + registered id
- Type.NAMED_EXT + registered namespace+typename
struct:
- Type.STRUCT + struct meta
- Type.NAMED_STRUCT + struct meta

Every type must be registered with an ID or name first. The registration can be used for security check and type identification.

Struct is a special type, depending whether schema compatibility is enabled, Fury will write struct meta differently.

Struct Schema consistent

If schema consistent mode is enabled globally when creating fury, type meta will be written as a fury unsigned varint of type_id. Schema evolution related meta will be ignored.
If schema evolution mode is enabled globally when creating fury, and current class is configured to use schema consistent mode like struct vs table in flatbuffers:
- Type meta will be add to captured_type_defs: captured_type_defs[type def stub] = map size ahead when registering type.
- Get index of the meta in captured_type_defs, write that index as | unsigned varint: index |.

Struct Schema evolution

If schema evolution mode is enabled globally when creating fury, and enabled for current type, type meta will be written using one of the following mode. Which mode to use is configured when creating fury.

Normal mode(meta share not enabled):
- If type meta hasn't been written before, add type def to captured_type_defs: captured_type_defs[type def] = map size.
- Get index of the meta in captured_type_defs, write that index as | unsigned varint: index |.
- After finished the serialization of the object graph, fury will start to write captured_type_defs:
  - Firstly, set current to meta start offset of fury header
  - Then write captured_type_defs one by one:
    buffer.write_var_uint32(len(writting_type_defs) - len(schema_consistent_type_def_stubs)) for type_meta in writting_type_defs: if not type_meta.is_stub(): type_meta.write_type_def(buffer) writing_type_defs = copy(schema_consistent_type_def_stubs)

Meta share mode: the writing steps are same as the normal mode, but captured_type_defs will be shared across multiple serializations of different objects. For example, suppose we have a batch to serialize:

captured_type_defs = {}
stream = ...
# add `Type1` to `captured_type_defs` and write `Type1`
fury.serialize(stream, [Type1()])
# add `Type2` to `captured_type_defs` and write `Type2`, `Type1` is written before.
fury.serialize(stream, [Type1(), Type2()])
# `Type1` and `Type2` are written before, no need to write meta.
fury.serialize(stream, [Type1(), Type2()])

Streaming mode(streaming mode doesn't support meta share):
- If type meta hasn't been written before, the data will be written as:
```
| unsigned varint: 0b11111111 | type def |
```
- If type meta has been written before, the data will be written as:
```
| unsigned varint: written index << 1 |
```
  written index is the id in captured_type_defs.
- With this mode, meta start offset can be omitted.

The normal mode and meta share mode will forbid streaming writing since it needs to look back for update the start offset after the whole object graph writing and meta collecting is finished. Only in this way we can ensure deserialization failure in meta share mode doesn't lost shared meta.

Type Def

Here we mainly describe the meta layout for schema evolution mode:

|    8 bytes header    |   variable bytes   |  variable bytes   |
+----------------------+--------------------+-------------------+
| global binary header |    meta header     |    fields meta    |

For languages which support inheritance, if parent class and subclass has fields with same name, using field in subclass.

Global binary header

50 bits hash + 1bit compress flag + write fields meta + 12 bits meta size. Right is the lower bits.

lower 12 bits are used to encode meta size. If meta size >= 0b111_1111_1111, then write meta_ size - 0b111_1111_1111 next.
13rd bit is used to indicate whether to write fields meta. When this class is schema-consistent or use registered serializer, fields meta will be skipped. Class Meta will be used for share namespace + type name only.
14rd bit is used to indicate whether meta is compressed.
Other 50 bits is used to store the unique hash of flags + all layers class meta.

Meta header

Meta header is a 8 bits number value.

Lowest 5 digits 0b00000~0b11110 are used to record num fields. 0b11111 is preserved to indicate that Fury need to read more bytes for length using Fury unsigned int encoding. Note that num_fields is the number of compatible fields. Users can use tag id to mark some fields as compatible fields in schema consistent context. In such cases, schema consistent fields will be serialized first, then compatible fields will be serialized next. At deserialization, Fury will use fields info of those fields which aren't annotated by tag id for deserializing schema consistent fields, then use fields info in meta for deserializing compatible fields.
The 6th bit: 0 for registered by id, 1 for registered by name.
Remaining 2 bits are reserved for future extension.

Fields meta

Format:

|   field info: variable bytes    | variable bytes  | ... |
+---------------------------------+-----------------+-----+
| header + type info + field name | next field info | ... |

Field Header

Field Header is 8 bits, annotation can be used to provide more specific info. If annotation not exists, fury will infer those info automatically.

The format for field header is:

2 bits field name encoding + 4 bits size + nullability flag + ref tracking flag

Detailed spec:

2 bits field name encoding:
- encoding: UTF8/ALL_TO_LOWER_SPECIAL/LOWER_UPPER_DIGIT_SPECIAL/TAG_ID
- If tag id is used, field name will be written by an unsigned varint tag id, and 2 bits encoding will be 11.
size of field name:
- The 4 bits size: 0~14 will be used to indicate length 1~15, the value 15 indicates to read more bytes, the encoding will encode size - 15 as a varint next.
- If encoding is TAG_ID, then num_bytes of field name will be used to store tag id.
ref tracking: when set to 1, ref tracking will be enabled for this field.
nullability: when set to 1, this field can be null.

Field Type Info

Field type info is written as unsigned int8. Detailed id spec is:

For struct registered by id, it will be Type.STRUCT.
For struct registered by name, it will be Type.NAMED_STRUCT.
For enum registered by id, it will be Type.ENUM.
For enum registered by name, it will be Type.NAMED_ENUM.
For ext type registered by id, it will be Type.EXT.
For ext type registered by name, it will be Type.NAMED_EXT.
For list/set type, it will be written as Type.LIST/SET, then write element type recursively.
For 1D primitive array type, it will be written as Type.XXX_ARRAY.
For multi-dimensional primitive array type with same size on each dim, it will be written as Type.TENSOR.
For other array type, it will be written as Type.LIST, then write element type recursively.
For map type, it will be written as Type.MAP, then write key and value type recursively.
For other types supported by fury directly, it will be fury type id for that type.
For other types not determined at compile time, write Type.UNKNOWN instead. For such types, actual type will be written when serializing such field values.

Polymorphism spec:

struct/named_struct/ext/named_ext are taken as polymorphic, the meta for those types are written separately instead of inlining here to reduce meta space cost if object of this type is serialized in current object graph multiple times, and the field value may be null too.
enum is taken as morphic, if deserialization doesn't have this field, or the type is not enum, enum value will be skipped.
list/map/set are taken as morphic, when serializing values of those type, the concrete types won't be written again.
Other types that fury supported are taken as morphic too.

List/Set/Map nested type spec:

list: | list type id | nested type id << 2 + nullability flag + ref tracking flag | ... multi-layer type info |
set: | set type id | nested type id << 2 + nullability flag + ref tracking flag | ... multi-layer type info |
map: | set type id | key type info | value type info |
- Key type format: | nested type id << 2 + nullability flag + ref tracking flag | ... multi-layer type info |
- Value type format: | nested type id << 2 + nullability flag + ref tracking flag | ... multi-layer type info |

Field Name

If tag id is set, tag id will be used instead. Otherwise meta string of field name will be written instead.

Field order

Field order are left as implementation details, which is not exposed to specification, the deserialization need to resort fields based on Fury fields sort algorithms. In this way, fury can compute statistics for field names or types and using a more compact encoding.

Extended Type Meta with Inheritance support

If one want to support inheritance for struct, one can implement following spec.

Schema consistent

Fields are serialized from parent type to leaf type. Fields are sorted using fury struct fields sort algorithms.

Schema Evolution

Meta layout for schema evolution mode:

|    8 bytes header    | variable bytes | variable bytes |   variable bytes   |   variable bytes   |
+----------------------+----------------+----------------+--------------------+--------------------+
| global binary header |  meta header   |  fields meta   | parent meta header | parent fields meta |

Meta header

Meta header is a 64 bits number value encoded in little endian order.

Lowest 4 digits 0b0000~0b1110 are used to record num classes. 0b1111 is preserved to indicate that Fury need to read more bytes for length using Fury unsigned int encoding. If current type doesn't has parent type, or parent type doesn't have fields to serialize, or we're in a context which serialize fields of current type only, num classes will be 1.
The 5th bit is used to indicate whether this type needs schema evolution.
Other 56 bits are used to store the unique hash of flags + all layers type meta.

Single layer type meta

| unsigned varint | var uint |  field info: variable bytes   | variable bytes  | ... |
+-----------------+----------+-------------------------------+-----------------+-----+
|   num_fields    | type id  | header + type id + field name | next field info | ... |

Other layers type meta

Same encoding algorithm as the previous layer.

Meta String

Meta string is mainly used to encode meta strings such as field names.

Encoding Algorithms

String binary encoding algorithm:

Algorithm	Pattern	Description
LOWER_SPECIAL	`a-z._$\|`	every char is written using 5 bits, `a-z`: `0b00000~0b11001`, `._$\|`: `0b11010~0b11101`, prepend one bit at the start to indicate whether strip last char since last byte may have 7 redundant bits(1 indicates strip last char)
LOWER_UPPER_DIGIT_SPECIAL	`a-zA-Z0~9._`	every char is written using 6 bits, `a-z`: `0b00000~0b11001`, `A-Z`: `0b11010~0b110011`, `0~9`: `0b110100~0b111101`, `._`: `0b111110~0b111111`, prepend one bit at the start to indicate whether strip last char since last byte may have 7 redundant bits(1 indicates strip last char)
UTF-8	any chars	UTF-8 encoding

Encoding flags:

Encoding Flag	Pattern	Encoding Algorithm
LOWER_SPECIAL	every char is in `a-z._\|`	`LOWER_SPECIAL`
FIRST_TO_LOWER_SPECIAL	every char is in `a-z._` except first char is upper case	replace first upper case char to lower case, then use `LOWER_SPECIAL`
ALL_TO_LOWER_SPECIAL	every char is in `a-zA-Z._`	replace every upper case char by `\|` + `lower case`, then use `LOWER_SPECIAL`, use this encoding if it's smaller than Encoding `LOWER_UPPER_DIGIT_SPECIAL`
LOWER_UPPER_DIGIT_SPECIAL	every char is in `a-zA-Z._`	use `LOWER_UPPER_DIGIT_SPECIAL` encoding if it's smaller than Encoding `FIRST_TO_LOWER_SPECIAL`
UTF8	any utf-8 char	use `UTF-8` encoding
Compression	any utf-8 char	lossless compression

Notes:

Depending on cases, one can choose encoding flags + data jointly, uses 3 bits of first byte for flags and other bytes for data.

Value Format

Basic types

bool

size: 1 byte
format: 0 for false, 1 for true

int8

size: 1 byte
format: write as pure byte.

int16

size: 2 byte
byte order: raw bytes of little endian order

unsigned int32

size: 4 byte
byte order: raw bytes of little endian order

unsigned varint32

size: 1~5 byte
Format: The most significant bit (MSB) in every byte indicates whether to have the next byte. If first bit is set i.e. b & 0x80 == 0x80, then the next byte should be read until the first bit of the next byte is unset.

signed int32

size: 4 byte
byte order: raw bytes of little endian order

signed varint32

size: 1~5 byte
Format: First convert the number into positive unsigned int by (v << 1) ^ (v >> 31) ZigZag algorithm, then encode it as an unsigned varint.

unsigned int64

size: 8 byte
byte order: raw bytes of little endian order

unsigned varint64

size: 1~9 byte
Fury SLI(Small long as int) Encoding:
- If long is in [0, 2147483647], encode as 4 bytes int: | little-endian: ((int) value) << 1 |
- Otherwise write as 9 bytes: | 0b1 | little-endian 8 bytes long |
Fury PVL(Progressive Variable-length Long) Encoding:
- positive long format: first bit in every byte indicates whether to have the next byte. If first bit is set i.e. b & 0x80 == 0x80, then the next byte should be read until the first bit is unset.

signed int64

size: 8 byte
byte order: raw bytes of little endian order

signed varint64

size: 1~9 byte
Fury SLI(Small long as int) Encoding:
- If long is in [-1073741824, 1073741823], encode as 4 bytes int: | little-endian: ((int) value) << 1 |
- Otherwise write as 9 bytes: | 0b1 | little-endian 8 bytes long |
Fury PVL(Progressive Variable-length Long) Encoding:
- First convert the number into positive unsigned long by (v << 1) ^ (v >> 63) ZigZag algorithm to reduce cost of small negative numbers, then encoding it as an unsigned long.

float32

size: 4 byte
format: encode the specified floating-point value according to the IEEE 754 floating-point "single format" bit layout, preserving Not-a-Number (NaN) values, then write as binary by little endian order.

float64

size: 8 byte
format: encode the specified floating-point value according to the IEEE 754 floating-point "double format" bit layout, preserving Not-a-Number (NaN) values. then write as binary by little endian order.

string

Format:

| unsigned varint64: size << 2 `bitor` 2 bits encoding flags | binary data |

size + encoding will be concat as a long and encoded as an unsigned varint64. The little 2 bits is used for encoding: 0 for latin1(ISO-8859-1), 1 for utf-16, 2 for utf-8.
encoded string binary data based on encoding: latin/utf-16/utf-8.

Which encoding to choose:

For JDK8: fury detect latin at runtime, if string is latin string, then use latin encoding, otherwise use utf-16.
For JDK9+: fury use coder in String object for encoding, latin/utf-16 will be used for encoding.
If the string is encoded by utf-8, then fury will use utf-8 to decode the data. Cross-language string serialization of fury uses utf-8 by default.

list

Format:

| unsigned varint64: length | 1 byte elements header | elements data |

elements header

In most cases, all elements are same type and not null, elements header will encode those homogeneous information to avoid the cost of writing it for every element. Specifically, there are four kinds of information which will be encoded by elements header, each use one bit:

If track elements ref, use the first bit 0b1 of the header to flag it.
If the elements have null, use the second bit 0b10 of the header to flag it. If ref tracking is enabled for this element type, this flag is invalid.
If the element types are not the declared type, use the 3rd bit 0b100 of the header to flag it.
If the element types are different, use the 4rd bit 0b1000 header to flag it.

By default, all bits are unset, which means all elements won't track ref, all elements are same type, not null and the actual element is the declared type in the custom type field.

The implementation can generate different deserialization code based read header, and look up the generated code from a linear map/list.

elements data

Based on the elements header, the serialization of elements data may skip ref flag/null flag/element type info.

fury = ...
buffer = ...
elems = ...
if element_type_is_same:
    if not is_declared_type:
        fury.write_type(buffer, elem_type)
    elem_serializer = get_serializer(...)
    if track_ref:
        for elem in elems:
            if not ref_resolver.write_ref_or_null(buffer, elem):
                elem_serializer.write(buffer, elem)
    elif has_null:
        for elem in elems:
            if elem is None:
                buffer.write_byte(null_flag)
            else:
                buffer.write_byte(not_null_flag)
                elem_serializer.write(buffer, elem)
    else:
        for elem in elems:
            elem_serializer.write(buffer, elem)
else:
    if track_ref:
        for elem in elems:
            fury.write_ref(buffer, elem)
    elif has_null:
        for elem in elems:
            fury.write_nullable(buffer, elem)
    else:
        for elem in elems:
            fury.write_value(buffer, elem)

CollectionSerializer#writeElements can be taken as an example.

array

primitive array

Primitive array are taken as a binary buffer, serialization will just write the length of array size as an unsigned int, then copy the whole buffer into the stream.

Such serialization won't compress the array. If users want to compress primitive array, users need to register custom serializers for such types or mark it as list type.

Tensor

Tensor is a special primitive multi-dimensional array which all dimensions have same size and type. The serialization format is:

| num_dims(unsigned varint) | shape[0](unsigned varint) | shape[...] | shape[N] | element type | data |

The data is continuous to reduce copy and may zero-copy in some cases.

object array

Object array is serialized using the list format. Object component type will be taken as list element generic type.

map

Map uses a chunk by chunk based Format:

| length(unsigned varint) | key value chunk data | ... | key value chunk data |

map key-value chunk data

Map iteration is too expensive, Fury won't compute the header like for list since it introduce considerable overhead. Users can use MapFieldInfo annotation to provide the header in advance. Otherwise Fury will use first key-value pair to predict header optimistically, and update the chunk header if the prediction failed at some pair.

Fury will serialize the map chunk by chunk, every chunk has 255 pairs at most.

|    1 byte      |     1 byte     | variable bytes  |
+----------------+----------------+-----------------+
|    KV header   | chunk size: N  |   N*2 objects   |

KV header:

If track key ref, use the first bit 0b1 of the header to flag it.
If the key has null, use the second bit 0b10 of the header to flag it. If ref tracking is enabled for this key type, this flag is invalid.
If the actual key type of map is not the declared key type, use the 3rd bit 0b100 of the header to flag it.
If track value ref, use the 4th bit 0b1000 of the header to flag it.
If the value has null, use the 5th bit 0b10000 of the header to flag it. If ref tracking is enabled for this value type, this flag is invalid.
If the value type of map is not the declared value type, use the 6rd bit 0b100000 of the header to flag it.
If key or value is null, that key and value will be written as a separate chunk, and chunk size writing will be skipped too.

If streaming write is enabled, which means Fury can't update written chunk size. In such cases, map key-value data format will be:

|    1 byte      | variable bytes  |
+----------------+-----------------+
|    KV header   |   N*2 objects   |

KV header will be a header marked by MapFieldInfo in java. For languages such as golang, this can be computed in advance for non-interface types most times. The implementation can generate different deserialization code based read header, and look up the generated code from a linear map/list.

Why serialize chunk by chunk?

When fury will use first key-value pair to predict header optimistically, it can't know how many pairs have same meta(tracking kef ref, key has null and so on). If we don't write chunk by chunk with max chunk size, we must write at least X bytes to take up a place for later to update the number which has same elements, X is the num_bytes for encoding varint encoding of map size.

And most map size are smaller than 255, if all pairs have same data, the chunk will be 1. This is common in golang/rust, which object are not reference by default.

Also, if only one or two keys have different meta, we can make it into a different chunk, so that most pairs can share meta.

The implementation can accumulate read count with map size to decide whether to read more chunks.

enum

Enums are serialized as an unsigned var int. If the order of enum values change, the deserialized enum value may not be the value users expect. In such cases, users must register enum serializer by make it write enum value as an enumerated string with unique hash disabled.

decimal

Not supported for now.

struct

Struct means object of class/pojo/struct/bean/record type. Struct will be serialized by writing its fields data in fury order.

Depending on schema compatibility, structs will have different formats.

field order

Field will be ordered as following, every group of fields will have its own order:

primitive fields:
- larger size type first, smaller later, variable size type last.
- when same size, sort by type id
- when same size and type id, sort by snake case field name
- types: bool/int8/int16/int32/varint32/int64/varint64/sliint64/float16/float32/float64
nullable primitive fields: same order as primitive fields
morphic fields: same type together, then sorted by field name lexicographically using snake case style.
unknown fields: same sort algorithms as morphic fields
list fields: same sort algorithms as morphic fields
set fields: same sort algorithms as morphic fields
map fields: same sort algorithms as morphic fields

Field order

Fields in a struct are sorted in a ascending order by:

primitive fields first: bool/int8/int16/int32/varint32/int64/varint64/sliint64/float16/float32/float64, sorted by type id.
nullable primitive fields
morphic types except list/set/map
unknown types
list types
set types
map types

If two fields have same type, then sort by snake_case styled field name.

schema consistent

Object will be written as:

|    4 byte     |  variable bytes  |
+---------------+------------------+
|   type hash   |   field values   |

Type hash is used to check the type schema consistency across languages. Type hash will be the first 32 bits of 56 bits value of the type meta.

Object fields will be serialized one by one using following format:

not null primitive field value:
|   var bytes    |
+----------------+
|   value data   |
+----------------+
nullable primitive field value:
| one byte  |   var bytes   |
+-----------+---------------+
| null flag |  field value  |
+-----------+---------------+
field value of final type with ref tracking:
| var bytes | var objects |
+-----------+-------------+
| ref meta  | value data  |
+-----------+-------------+
field value of final type without ref tracking:
| one byte  | var objects |
+-----------+-------------+
| null flag | field value |
+-----------+-------------+
field value of non-final type with ref tracking:
| one byte  | var bytes | var objects |
+-----------+-------------+-------------+
| ref meta  | type meta  | value data  |
+-----------+-------------+-------------+
field value of non-final type without ref tracking:
| one byte  | var bytes | var objects |
+-----------+------------+------------+
| null flag | type meta | value data |
+-----------+------------+------------+

Schema evolution

Schema evolution have similar format as schema consistent mode for object except:

For the object type, schema consistent mode will write type by id only, but schema evolution mode will write type consisting of field names, types and other meta too, see Type meta.
Type meta of final custom type needs to be written too, because peers may not have this type defined.

Type

Type will be serialized using type meta format.

Implementation guidelines

How to reduce memory read/write code

Try to merge multiple bytes into an int/long write before writing to reduce memory IO and bound check cost.
Read multiple bytes as an int/long, then split into multiple bytes to reduce memory IO and bound check cost.
Try to use one varint/long to write flags and length together to save one byte cost and reduce memory io.
Condition branches are less expensive compared to memory IO cost unless there are too many branches.

Fast deserialization for static languages without runtime codegen support

For type evolution, the serializer will encode the type meta into the serialized data. The deserializer will compare this meta with class meta in the current process, and use the diff to determine how to deserialize the data.

For java/javascript/python, we can use the diff to generate serializer code at runtime and load it as class/function for deserialization. In this way, the type evolution will be as fast as type consist mode.

For C++/Rust, we can't generate the serializer code at runtime. So we need to generate the code at compile-time using meta programming. But at that time, we don't know the type schema in other processes, so we can't generate the serializer code for such inconsistent types. We may need to generate the code which has a loop and compare field name one by one to decide whether to deserialize and assign the field or skip the field value.

One fast way is that we can optimize the string comparison into jump instructions:

Assume the current type has n fields, and the peer type has n1 fields.
Generate an auto growing field id from 0 for every sorted field in the current type at the compile time.
Compare the received type meta with current type, generate same id if the field name is same, otherwise generate an auto growing id starting from n, cache this meta at runtime.
Iterate the fields of received type meta, use a switch to compare the field id to deserialize data and assign/skip field value. Continuous field id will be optimized into jump in switch block, so it will very fast.

Here is an example, suppose process A has a class Foo with version 1 defined as Foo1, process B has a class Foo with version 2 defined as Foo2:

// class Foo with version 1
class Foo1 {
  int32_t v1; // id 0
  std::string v2; // id 1
};
// class Foo with version 2
class Foo2 {
  // id 0, but will have id 2 in process A
  bool v0;
  // id 1, but will have id 0 in process A
  int32_t v1;
  // id 2, but will have id 3 in process A
  int64_t long_value;
  // id 3, but will have id 1 in process A
  std::string v2;
  // id 4, but will have id 4 in process A
  std::vector<std::string> list;
};

When process A received serialized Foo2 from process B, here is how it deserialize the data:

Foo1 foo1 = ...;
const std::vector<fury::FieldInfo> &field_infos = type_meta.field_infos;
for (const auto &field_info : field_infos) {
  switch (field_info.field_id) {
    case 0:
      foo1.v1 = buffer.read_varint32();
      break;
    case 1:
      foo1.v2 = fury.read_string();
      break;
    default:
      fury.skip_data(field_info);
  }
}

Cross-language Serialization Specification​

Type Systems​

Data Types​

Polymorphisms​

Type disambiguation​

Type ID​

Type mapping​

Spec overview​

Fury header​

Reference Meta​

Type Meta​

Struct Schema consistent​

Struct Schema evolution​

Type Def​

Global binary header​

Meta header​

Fields meta​

Field Header​

Field Type Info​

Field Name​

Field order​

Extended Type Meta with Inheritance support​

Schema consistent​

Schema Evolution​

Meta header​

Single layer type meta​

Other layers type meta​

Meta String​

Encoding Algorithms​

Value Format​

Basic types​

bool​

int8​

int16​

unsigned int32​

unsigned varint32​

signed int32​

signed varint32​

unsigned int64​

unsigned varint64​

signed int64​

signed varint64​

float32​

float64​

string​

list​

elements header​

elements data​

array​

primitive array​

Tensor​

object array​

map​

map key-value chunk data​

Why serialize chunk by chunk?​

enum​

decimal​

struct​

field order​

Field order​

schema consistent​

Schema evolution​

Type​

Implementation guidelines​

How to reduce memory read/write code​

Fast deserialization for static languages without runtime codegen support​

Cross-language Serialization Specification

Type Systems

Data Types

Polymorphisms

Type disambiguation

Type ID

Type mapping

Spec overview

Fury header

Reference Meta

Type Meta

Struct Schema consistent

Struct Schema evolution

Type Def

Global binary header

Meta header

Fields meta

Field Header

Field Type Info

Field Name

Field order

Extended Type Meta with Inheritance support

Schema consistent

Schema Evolution

Meta header

Single layer type meta

Other layers type meta

Meta String

Encoding Algorithms

Value Format

Basic types

bool

int8

int16

unsigned int32

unsigned varint32

signed int32

signed varint32

unsigned int64

unsigned varint64

signed int64

signed varint64

float32

float64

string

list

elements header

elements data

array

primitive array

Tensor

object array

map

map key-value chunk data

Why serialize chunk by chunk?

enum

decimal

struct

field order

Field order

schema consistent

Schema evolution

Type

Implementation guidelines

How to reduce memory read/write code

Fast deserialization for static languages without runtime codegen support