Run Format

Source file src/pkg/encoding/gob/doc.go

     1	// Copyright 2009 The Go Authors. All rights reserved.
     2	// Use of this source code is governed by a BSD-style
     3	// license that can be found in the LICENSE file.
     4	
     5	/*
     6	Package gob manages streams of gobs - binary values exchanged between an
     7	Encoder (transmitter) and a Decoder (receiver).  A typical use is transporting
     8	arguments and results of remote procedure calls (RPCs) such as those provided by
     9	package "rpc".
    10	
    11	The implementation compiles a custom codec for each data type in the stream and
    12	is most efficient when a single Encoder is used to transmit a stream of values,
    13	amortizing the cost of compilation.
    14	
    15	Basics
    16	
    17	A stream of gobs is self-describing.  Each data item in the stream is preceded by
    18	a specification of its type, expressed in terms of a small set of predefined
    19	types.  Pointers are not transmitted, but the things they point to are
    20	transmitted; that is, the values are flattened.  Recursive types work fine, but
    21	recursive values (data with cycles) are problematic.  This may change.
    22	
    23	To use gobs, create an Encoder and present it with a series of data items as
    24	values or addresses that can be dereferenced to values.  The Encoder makes sure
    25	all type information is sent before it is needed.  At the receive side, a
    26	Decoder retrieves values from the encoded stream and unpacks them into local
    27	variables.
    28	
    29	Types and Values
    30	
    31	The source and destination values/types need not correspond exactly.  For structs,
    32	fields (identified by name) that are in the source but absent from the receiving
    33	variable will be ignored.  Fields that are in the receiving variable but missing
    34	from the transmitted type or value will be ignored in the destination.  If a field
    35	with the same name is present in both, their types must be compatible. Both the
    36	receiver and transmitter will do all necessary indirection and dereferencing to
    37	convert between gobs and actual Go values.  For instance, a gob type that is
    38	schematically,
    39	
    40		struct { A, B int }
    41	
    42	can be sent from or received into any of these Go types:
    43	
    44		struct { A, B int }	// the same
    45		*struct { A, B int }	// extra indirection of the struct
    46		struct { *A, **B int }	// extra indirection of the fields
    47		struct { A, B int64 }	// different concrete value type; see below
    48	
    49	It may also be received into any of these:
    50	
    51		struct { A, B int }	// the same
    52		struct { B, A int }	// ordering doesn't matter; matching is by name
    53		struct { A, B, C int }	// extra field (C) ignored
    54		struct { B int }	// missing field (A) ignored; data will be dropped
    55		struct { B, C int }	// missing field (A) ignored; extra field (C) ignored.
    56	
    57	Attempting to receive into these types will draw a decode error:
    58	
    59		struct { A int; B uint }	// change of signedness for B
    60		struct { A int; B float }	// change of type for B
    61		struct { }			// no field names in common
    62		struct { C, D int }		// no field names in common
    63	
    64	Integers are transmitted two ways: arbitrary precision signed integers or
    65	arbitrary precision unsigned integers.  There is no int8, int16 etc.
    66	discrimination in the gob format; there are only signed and unsigned integers.  As
    67	described below, the transmitter sends the value in a variable-length encoding;
    68	the receiver accepts the value and stores it in the destination variable.
    69	Floating-point numbers are always sent using IEEE-754 64-bit precision (see
    70	below).
    71	
    72	Signed integers may be received into any signed integer variable: int, int16, etc.;
    73	unsigned integers may be received into any unsigned integer variable; and floating
    74	point values may be received into any floating point variable.  However,
    75	the destination variable must be able to represent the value or the decode
    76	operation will fail.
    77	
    78	Structs, arrays and slices are also supported. Structs encode and decode only
    79	exported fields. Strings and arrays of bytes are supported with a special,
    80	efficient representation (see below). When a slice is decoded, if the existing
    81	slice has capacity the slice will be extended in place; if not, a new array is
    82	allocated. Regardless, the length of the resulting slice reports the number of
    83	elements decoded.
    84	
    85	Functions and channels will not be sent in a gob. Attempting to encode such a value
    86	at top the level will fail. A struct field of chan or func type is treated exactly
    87	like an unexported field and is ignored.
    88	
    89	Gob can encode a value of any type implementing the GobEncoder or
    90	encoding.BinaryMarshaler interfaces by calling the corresponding method,
    91	in that order of preference.
    92	
    93	Gob can decode a value of any type implementing the GobDecoder or
    94	encoding.BinaryUnmarshaler interfaces by calling the corresponding method,
    95	again in that order of preference.
    96	
    97	Encoding Details
    98	
    99	This section documents the encoding, details that are not important for most
   100	users. Details are presented bottom-up.
   101	
   102	An unsigned integer is sent one of two ways.  If it is less than 128, it is sent
   103	as a byte with that value.  Otherwise it is sent as a minimal-length big-endian
   104	(high byte first) byte stream holding the value, preceded by one byte holding the
   105	byte count, negated.  Thus 0 is transmitted as (00), 7 is transmitted as (07) and
   106	256 is transmitted as (FE 01 00).
   107	
   108	A boolean is encoded within an unsigned integer: 0 for false, 1 for true.
   109	
   110	A signed integer, i, is encoded within an unsigned integer, u.  Within u, bits 1
   111	upward contain the value; bit 0 says whether they should be complemented upon
   112	receipt.  The encode algorithm looks like this:
   113	
   114		uint u;
   115		if i < 0 {
   116			u = (^i << 1) | 1	// complement i, bit 0 is 1
   117		} else {
   118			u = (i << 1)	// do not complement i, bit 0 is 0
   119		}
   120		encodeUnsigned(u)
   121	
   122	The low bit is therefore analogous to a sign bit, but making it the complement bit
   123	instead guarantees that the largest negative integer is not a special case.  For
   124	example, -129=^128=(^256>>1) encodes as (FE 01 01).
   125	
   126	Floating-point numbers are always sent as a representation of a float64 value.
   127	That value is converted to a uint64 using math.Float64bits.  The uint64 is then
   128	byte-reversed and sent as a regular unsigned integer.  The byte-reversal means the
   129	exponent and high-precision part of the mantissa go first.  Since the low bits are
   130	often zero, this can save encoding bytes.  For instance, 17.0 is encoded in only
   131	three bytes (FE 31 40).
   132	
   133	Strings and slices of bytes are sent as an unsigned count followed by that many
   134	uninterpreted bytes of the value.
   135	
   136	All other slices and arrays are sent as an unsigned count followed by that many
   137	elements using the standard gob encoding for their type, recursively.
   138	
   139	Maps are sent as an unsigned count followed by that many key, element
   140	pairs. Empty but non-nil maps are sent, so if the sender has allocated
   141	a map, the receiver will allocate a map even if no elements are
   142	transmitted.
   143	
   144	Structs are sent as a sequence of (field number, field value) pairs.  The field
   145	value is sent using the standard gob encoding for its type, recursively.  If a
   146	field has the zero value for its type, it is omitted from the transmission.  The
   147	field number is defined by the type of the encoded struct: the first field of the
   148	encoded type is field 0, the second is field 1, etc.  When encoding a value, the
   149	field numbers are delta encoded for efficiency and the fields are always sent in
   150	order of increasing field number; the deltas are therefore unsigned.  The
   151	initialization for the delta encoding sets the field number to -1, so an unsigned
   152	integer field 0 with value 7 is transmitted as unsigned delta = 1, unsigned value
   153	= 7 or (01 07).  Finally, after all the fields have been sent a terminating mark
   154	denotes the end of the struct.  That mark is a delta=0 value, which has
   155	representation (00).
   156	
   157	Interface types are not checked for compatibility; all interface types are
   158	treated, for transmission, as members of a single "interface" type, analogous to
   159	int or []byte - in effect they're all treated as interface{}.  Interface values
   160	are transmitted as a string identifying the concrete type being sent (a name
   161	that must be pre-defined by calling Register), followed by a byte count of the
   162	length of the following data (so the value can be skipped if it cannot be
   163	stored), followed by the usual encoding of concrete (dynamic) value stored in
   164	the interface value.  (A nil interface value is identified by the empty string
   165	and transmits no value.) Upon receipt, the decoder verifies that the unpacked
   166	concrete item satisfies the interface of the receiving variable.
   167	
   168	The representation of types is described below.  When a type is defined on a given
   169	connection between an Encoder and Decoder, it is assigned a signed integer type
   170	id.  When Encoder.Encode(v) is called, it makes sure there is an id assigned for
   171	the type of v and all its elements and then it sends the pair (typeid, encoded-v)
   172	where typeid is the type id of the encoded type of v and encoded-v is the gob
   173	encoding of the value v.
   174	
   175	To define a type, the encoder chooses an unused, positive type id and sends the
   176	pair (-type id, encoded-type) where encoded-type is the gob encoding of a wireType
   177	description, constructed from these types:
   178	
   179		type wireType struct {
   180			ArrayT  *ArrayType
   181			SliceT  *SliceType
   182			StructT *StructType
   183			MapT    *MapType
   184		}
   185		type arrayType struct {
   186			CommonType
   187			Elem typeId
   188			Len  int
   189		}
   190		type CommonType struct {
   191			Name string // the name of the struct type
   192			Id  int    // the id of the type, repeated so it's inside the type
   193		}
   194		type sliceType struct {
   195			CommonType
   196			Elem typeId
   197		}
   198		type structType struct {
   199			CommonType
   200			Field []*fieldType // the fields of the struct.
   201		}
   202		type fieldType struct {
   203			Name string // the name of the field.
   204			Id   int    // the type id of the field, which must be already defined
   205		}
   206		type mapType struct {
   207			CommonType
   208			Key  typeId
   209			Elem typeId
   210		}
   211	
   212	If there are nested type ids, the types for all inner type ids must be defined
   213	before the top-level type id is used to describe an encoded-v.
   214	
   215	For simplicity in setup, the connection is defined to understand these types a
   216	priori, as well as the basic gob types int, uint, etc.  Their ids are:
   217	
   218		bool        1
   219		int         2
   220		uint        3
   221		float       4
   222		[]byte      5
   223		string      6
   224		complex     7
   225		interface   8
   226		// gap for reserved ids.
   227		WireType    16
   228		ArrayType   17
   229		CommonType  18
   230		SliceType   19
   231		StructType  20
   232		FieldType   21
   233		// 22 is slice of fieldType.
   234		MapType     23
   235	
   236	Finally, each message created by a call to Encode is preceded by an encoded
   237	unsigned integer count of the number of bytes remaining in the message.  After
   238	the initial type name, interface values are wrapped the same way; in effect, the
   239	interface value acts like a recursive invocation of Encode.
   240	
   241	In summary, a gob stream looks like
   242	
   243		(byteCount (-type id, encoding of a wireType)* (type id, encoding of a value))*
   244	
   245	where * signifies zero or more repetitions and the type id of a value must
   246	be predefined or be defined before the value in the stream.
   247	
   248	See "Gobs of data" for a design discussion of the gob wire format:
   249	http://golang.org/doc/articles/gobs_of_data.html
   250	*/
   251	package gob
   252	
   253	/*
   254	Grammar:
   255	
   256	Tokens starting with a lower case letter are terminals; int(n)
   257	and uint(n) represent the signed/unsigned encodings of the value n.
   258	
   259	GobStream:
   260		DelimitedMessage*
   261	DelimitedMessage:
   262		uint(lengthOfMessage) Message
   263	Message:
   264		TypeSequence TypedValue
   265	TypeSequence
   266		(TypeDefinition DelimitedTypeDefinition*)?
   267	DelimitedTypeDefinition:
   268		uint(lengthOfTypeDefinition) TypeDefinition
   269	TypedValue:
   270		int(typeId) Value
   271	TypeDefinition:
   272		int(-typeId) encodingOfWireType
   273	Value:
   274		SingletonValue | StructValue
   275	SingletonValue:
   276		uint(0) FieldValue
   277	FieldValue:
   278		builtinValue | ArrayValue | MapValue | SliceValue | StructValue | InterfaceValue
   279	InterfaceValue:
   280		NilInterfaceValue | NonNilInterfaceValue
   281	NilInterfaceValue:
   282		uint(0)
   283	NonNilInterfaceValue:
   284		ConcreteTypeName TypeSequence InterfaceContents
   285	ConcreteTypeName:
   286		uint(lengthOfName) [already read=n] name
   287	InterfaceContents:
   288		int(concreteTypeId) DelimitedValue
   289	DelimitedValue:
   290		uint(length) Value
   291	ArrayValue:
   292		uint(n) FieldValue*n [n elements]
   293	MapValue:
   294		uint(n) (FieldValue FieldValue)*n  [n (key, value) pairs]
   295	SliceValue:
   296		uint(n) FieldValue*n [n elements]
   297	StructValue:
   298		(uint(fieldDelta) FieldValue)*
   299	*/
   300	
   301	/*
   302	For implementers and the curious, here is an encoded example.  Given
   303		type Point struct {X, Y int}
   304	and the value
   305		p := Point{22, 33}
   306	the bytes transmitted that encode p will be:
   307		1f ff 81 03 01 01 05 50 6f 69 6e 74 01 ff 82 00
   308		01 02 01 01 58 01 04 00 01 01 59 01 04 00 00 00
   309		07 ff 82 01 2c 01 42 00
   310	They are determined as follows.
   311	
   312	Since this is the first transmission of type Point, the type descriptor
   313	for Point itself must be sent before the value.  This is the first type
   314	we've sent on this Encoder, so it has type id 65 (0 through 64 are
   315	reserved).
   316	
   317		1f	// This item (a type descriptor) is 31 bytes long.
   318		ff 81	// The negative of the id for the type we're defining, -65.
   319			// This is one byte (indicated by FF = -1) followed by
   320			// ^-65<<1 | 1.  The low 1 bit signals to complement the
   321			// rest upon receipt.
   322	
   323		// Now we send a type descriptor, which is itself a struct (wireType).
   324		// The type of wireType itself is known (it's built in, as is the type of
   325		// all its components), so we just need to send a *value* of type wireType
   326		// that represents type "Point".
   327		// Here starts the encoding of that value.
   328		// Set the field number implicitly to -1; this is done at the beginning
   329		// of every struct, including nested structs.
   330		03	// Add 3 to field number; now 2 (wireType.structType; this is a struct).
   331			// structType starts with an embedded CommonType, which appears
   332			// as a regular structure here too.
   333		01	// add 1 to field number (now 0); start of embedded CommonType.
   334		01	// add 1 to field number (now 0, the name of the type)
   335		05	// string is (unsigned) 5 bytes long
   336		50 6f 69 6e 74	// wireType.structType.CommonType.name = "Point"
   337		01	// add 1 to field number (now 1, the id of the type)
   338		ff 82	// wireType.structType.CommonType._id = 65
   339		00	// end of embedded wiretype.structType.CommonType struct
   340		01	// add 1 to field number (now 1, the field array in wireType.structType)
   341		02	// There are two fields in the type (len(structType.field))
   342		01	// Start of first field structure; add 1 to get field number 0: field[0].name
   343		01	// 1 byte
   344		58	// structType.field[0].name = "X"
   345		01	// Add 1 to get field number 1: field[0].id
   346		04	// structType.field[0].typeId is 2 (signed int).
   347		00	// End of structType.field[0]; start structType.field[1]; set field number to -1.
   348		01	// Add 1 to get field number 0: field[1].name
   349		01	// 1 byte
   350		59	// structType.field[1].name = "Y"
   351		01	// Add 1 to get field number 1: field[1].id
   352		04	// struct.Type.field[1].typeId is 2 (signed int).
   353		00	// End of structType.field[1]; end of structType.field.
   354		00	// end of wireType.structType structure
   355		00	// end of wireType structure
   356	
   357	Now we can send the Point value.  Again the field number resets to -1:
   358	
   359		07	// this value is 7 bytes long
   360		ff 82	// the type number, 65 (1 byte (-FF) followed by 65<<1)
   361		01	// add one to field number, yielding field 0
   362		2c	// encoding of signed "22" (0x22 = 44 = 22<<1); Point.x = 22
   363		01	// add one to field number, yielding field 1
   364		42	// encoding of signed "33" (0x42 = 66 = 33<<1); Point.y = 33
   365		00	// end of structure
   366	
   367	The type encoding is long and fairly intricate but we send it only once.
   368	If p is transmitted a second time, the type is already known so the
   369	output will be just:
   370	
   371		07 ff 82 01 2c 01 42 00
   372	
   373	A single non-struct value at top level is transmitted like a field with
   374	delta tag 0.  For instance, a signed integer with value 3 presented as
   375	the argument to Encode will emit:
   376	
   377		03 04 00 06
   378	
   379	Which represents:
   380	
   381		03	// this value is 3 bytes long
   382		04	// the type number, 2, represents an integer
   383		00	// tag delta 0
   384		06	// value 3
   385	
   386	*/

View as plain text