Technical Specification

Version 1.0 – Release Candidate 1 – September 12, 2016

THIS DOCUMENT IS A RELEASE CANDIDATE FOR A PROPOSED FIX TECHNICAL STANDARD.

A RELEASE CANDIDATE HAS BEEN APPROVED BY THE GLOBAL TECHNICAL COMMITTEE AS AN

INITIAL STEP IN CREATING A NEW FIX TECHNICAL STANDARD. POTENTIAL ADOPTERS ARE

STRONGLY ENCOURAGED TO BEGIN WORKING WITH THE RELEASE CANDIDATE AND TO PROVIDE

FEEDBACK TO THE GLOBAL TECHNICAL COMMITTEE AND THE WORKING GROUP THAT SUBMITTED

THE PROPOSAL. THE FEEDBACK TO THE RELEASE CANDIDATE WILL DETERMINE IF ANOTHER

REVISION AND RELEASE CANDIDATE IS NECESSARY OR IF THE RELEASE CANDIDATE CAN BE

PROMOTED TO BECOME A FIX TECHNICAL STANDARD DRAFT.

Table of Contents

Introduction

Objective

The objective of this user guide is to support development of web applications that require FIX semantics. JSON is another alternative to existing FIX encodings, including tagvalue, FIXML, FAST, SBE and Google Protocol Buffers.

JSON encoding of FIX is optimized for operations in a web browser without the need for other software distribution to clients.

JSON encoding is also optimized for conversion to and from other FIX encodings. It is therefore a goal of this encoding not to discard any information that would be useful during such conversions.

It is not a goal to specify a particular template or schema format here, however it is a goal not to preclude the use of one.

On the scale of human-to-server web interactions, JSON provides acceptable performance. However, is not a goal to optimize JSON encoding to execute at very low latency like FIX binary encodings.

JavaScript and JSON

JavaScript is the predominant language for developing client-side logic in web applications. JavaScript is supported by all popular web browsers. A built-in feature of the language is JavaScript Object Notation (JSON). JSON syntax is identical to the code for creating JavaScript objects. It minimizes development effort while avoiding the complexity and greed for resources of XML parsers and the like.

An important characteristic of JSON is that it is self-describing. Elements have readable symbolic names. Therefore, message handlers can process message elements without looking up metadata in a data dictionary or schema.

JSON wire format can readily be parsed and encoded in other programming languages. There are several popular libraries for that purpose in Java, C#, C++, and so forth. These JSON implementations interoperate with JavaScript.

JSON Elements

JSON is so simple that the standard’s authors foresaw no need for versioning since it is never expected to change. To summarize:

  • An array is an ordered list of values.
  • An object is a collection of name/value pairs. Names are strings.
  • A value can be an object, array, number, string, boolean, or null.
  • Numbers are signed. There is no syntactic distinction between integer and floating point values. In practice, most implementations store all numbers as double precision binary floating point.
  • Strings are Unicode, with a few rules for escaping special values.
  • Booleans are true or false.
  • Arbitrary levels of object nesting are allowed.

Standards References

JavaScript is formalized as Standard ECMA-262 ECMAScript® 2015 Language Specification.

JSON is standardized by Standard ECMA-404 The JSON Data Interchange Format . The JSON standard is normative for this user guide.

Issues for Mapping JSON to FIX

This user guide provides standardized solutions to the following issues.

Dates and Times

JSON has no explicit provision for encoding dates or times. However, most languages/platforms that support JSON also support conversion of date/time to and from strings in ISO 8601 format (e.g. the JavaScript Date object).

Another potential issue is that particular languages/platforms only support millisecond precision (e.g. the JavaScript Date object), while FIX timestamps may require microsecond or nanosecond precision. Given that any timestamps captured on the client side are limited by PC clock precision, millisecond precision should be sufficient for web applications. When finer than millisecond precision is required, applications should adhere to the recommendations of FIX.5.0 SP2 EP206: Clock Synchronization Data Types Enhancements.

Decimal Representation

JSON does not provide a numeric data type that is suitable for storing prices, quantities, etc. Furthermore, since most implementations (including JavaScript) store all JSON numeric values using a binary floating point data type, attempts to represent decimals as scaled integers, as was done with FIX binary encodings, are unfortunately pointless.

Enumerations

Enumerations of valid values are needed for codes in FIX fields, but JSON has no special syntax for enumerations.

Languages/platforms that support JSON may also lack support for enumerations (e.g. JavaScript). Although it may be possible in such cases to emulate an enumeration with an associative array of symbolic names and values, deserialization of a code in JSON does not automatically associate to its symbolic name, and serialized strings or numbers are not constrained to valid values.

No tags

JSON does not have a built-in feature equivalent to FIX field tags or component IDs. This user guide suggests using symbolic names directly to avoid lookup by tag. The reduction in processing is offset by longer messages on the wire. This seems an acceptable trade-off since a web UI is unlikely to capture very large numbers of fields per message.

No Templates

JSON serialization and deserialization are not controlled by an external template or schema, only by an object that is being serialized. Each object is sui generis; JSON grew out of JavaScript, which does not have classes that objects must conform to, as realized by Java, C# and C++. It does have a prototype feature, but JavaScript objects are quite malleable. Properties and functions can be added on the fly. Nevertheless, it is possible to generate JSON objects corresponding to messages defined by the FIX Repository or FIX Orchestra.

Constraint

This user guide will depend only on standard JavaScript features that are implemented in most browsers and the JSON standard, without dependency on third-party frameworks.

Field Encoding

Data Types of Values

This user guide specifies that all of the semantic data types of FIX protocol should be mapped to JSON string values in order to maximize the ability of applications to simply display the data.

Application logic must be tailored to handle data with proper semantics if it does anything with the data other than simply display it. Since there is no feature to convey the semantic type in-band, the actual FIX type must be referenced at development time. It is recommended that applications use either FIX Repository or FIX Orchestra for this purpose.

JSON does not provide a numeric data type that is suitable for storing prices, quantities, etc. Furthermore, most implementations (e.g. JavaScript) store all JSON numeric values using a binary floating point data type. Using JSON string values to represent FIX protocol numeric types circumvents this issue, and allows applications to choose the most appropriate data type provided by their language/platform (e.g. the Java BigDecimal type).

Names

Names of fields must be encoded exactly as they are spelled and capitalized in the FIX Repository.

The field’s Name should be used instead of its Tag number to enable applications to display a human-readable form with little or no logic and without requiring a data dictionary in the browser. However, for user-defined fields, the Tag number may be used instead of the Name.

The field’s Value should be used instead of its SymbolicName since there are many cases (e.g. bonds, complex options) where the SymbolicName means something other than how it is being used.

Field Encoding

Fields are encoded in accordance with the JSON standard as name/value pairs. Values must be serialized as JSON strings.

Example of a FIX field encoded as a JSON name/value pair:

FIX tag=value Encoding JSON Encoding
31=47.50 “LastPx”: “47.50”
54=1 “Side”: “1”

Byte order

Since all JSON values, including numbers, are serialized as their string equivalent, there is no issue with byte order (endianness).

Message Structure

Field Presence

Although JSON does have a special value for null, it need not be used for a non-populated optional FIX field. Like FIX tag=value encoding, optional fields that are not populated are simply not serialized on the wire.

Field Order

Like FIX tag=value encoding, order of fields within a message or repeating group entry is not significant. All fields are accessed by name.

Message Framing

Each message is serialized as a JSON object, contained by opening and closing braces. A message may contain other JSON objects, specifically, repeating groups (see below).

Since this encoding is designed for use with web protocols, message framing is generally handled by the session layer protocol, e.g. HTTP or websockets. In these cases, no additional framing protocol is needed.

For cases where an additional framing protocol is needed, applications may use FIX Simple Open Framing Header (SOFH).

Header, Body, and Trailer

Every JSON message must have top-level fields named “Header”, “Body”, and “Trailer”:

{
    "Header": {},
    "Body": {},
    "Trailer": {}
}

This structure serves the goal of not discarding information that is useful when converting to/from other FIX encodings.

JSON encoding does not include a “CheckSum” field since it is unlikely to be useful at best, and likely to be incorrect at worst (e.g. if copied over from another FIX encoding).

Message Type

To identify a message on the wire, every JSON message should have a “MsgType” field in the “Header” sub-object. The value should be a valid Value of the “MsgType” field as defined in the FIX repository:

Example for a “NewOrderSingle” message:

{
    "Header": {
        "MsgType": "D"
    }
}

Repeating Groups

A repeating group is serialized as an array of JSON objects, each containing the fields that belong to a particular repeating group entry. A JSON array is surrounded by square brackets, and each entry is separated by a comma. Each entry is its own JSON object. Because some of the fields may be optional, not all entries are required to contain the same fields.

The count of entries is implicit to the array structure. There is no explicit NumInGroup field in the JSON encoding.

The name of a repeating group is the name of the associated NumInGroup field name as it appears in the FIX Repository.

Example of a “NoMDEntries” group with two entries:

{
    "Header": {
        "MsgType": "W"
    },
    "Body": {
        "NoMDEntries": [
            { "MDEntryType": "0", "MDEntryPx": "2179.75", "MDEntrySize": "175" },
            { "MDEntryType": "1", "MDEntryPx": "2180.25", "MDEntrySize": "125" }
        ]
    }
}

Empty group

An empty repeating group may be serialized in order to simplify conversion to/from other FIX encodings.

Nested groups

A JSON object for a repeating group entry may contain other objects to represent nested repeating groups.

Sample Messages

MarketDataSnapshotFullRefresh

{
    "Header": {
        "BeginString": "FIXT.1.1",
        "MsgType": "W",
        "MsgSeqNum": "4567",
        "SenderCompID": "SENDER",
        "TargetCompID": "TARGET",
        "SendingTime": "20160802-21:14:38.717"
    },
    "Body": {
        "SecurityIDSource": "8",
        "SecurityID": "ESU6",
        "MDReqID": "789",
        "NoMDEntries": [
            { "MDEntryType": "0", "MDEntryPx": "1.50", "MDEntrySize": "75", "MDEntryTime": "21:14:38.688" },
            { "MDEntryType": "1", "MDEntryPx": "1.75", "MDEntrySize": "25", "MDEntryTime": "21:14:38.688" }
        ]
    },
    "Trailer": {
    }
}

Unsupported Features

Metadata

This user guide only specifies wire format. There is no provision for external or internal metadata features.

Versioning

Explicit versioning is not supported by the JSON encoding of FIX. Generally, this is unnecessary since web sites serve client-side code all with page contents. Since the server controls encoding on both sides, they should always be internally consistent.