https://blog.trl.sn/blog/what-is-a-json-number/

🌈 Abstract

The article explores the concept of a JSON number, delving into the various JSON specifications and implementations to understand the nuances and practical implications of this seemingly straightforward data type.

🙋 Q&A

[01] Authoritative sources

1. What are the primary standards that define JSON? The two primary standards that define JSON are Ecma-404 and RFC 8259. These standards are semantically identical, but RFC 8259 provides additional recommendations for good interoperability.

2. What other related standards are mentioned? The article also mentions RFC 7493, which describes the closely related Internet JSON format, and JSON Schema, which defines a number data type that is normatively referenced by OpenAPI.

3. What does the ECMA-404 standard say about JSON numbers? According to ECMA-404, a JSON number is a sequence of decimal digits with an optional sign, fractional part, and exponent. The description is purely syntactical.

4. What does RFC 8259 say about JSON numbers? RFC 8259 provides an equivalent ABNF grammar to the ECMA-404 standard and explicitly allows implementations to set limits on the range and precision of numbers accepted. It also notes that good interoperability can be achieved by implementations that expect no more precision or range than what IEEE 754 binary64 (double precision) numbers provide.

5. What does RFC 7493 say about JSON numbers? RFC 7493 makes RFC 8259's informative note a normative "SHOULD NOT", stating that I-JSON messages SHOULD NOT include numbers that express greater magnitude or precision than an IEEE 754 double precision number provides. It recommends encoding numbers with greater range or precision as strings.

[02] JSON numbers in practice

1. What are the key points raised about JSON number implementations in different programming languages? The article examines the behavior of JSON number handling in several programming languages, including JavaScript, Python, C#, Java, Rust, and Go. It highlights the range and precision limitations of the default implementations, as well as the potential for precision loss when round-tripping certain numeric values.

2. What are the common patterns observed across the language implementations? The article summarizes the behavior of the various implementations when dealing with different types of numeric literals, such as small integers, large integers, high-precision decimals, and exponentials. It shows that the default behavior often involves using the built-in numeric types of the language, which can result in precision loss or errors for certain values.

3. What are the key takeaways from the language implementation analysis? The key takeaways are that JSON number handling can be complex, with different languages and libraries having varying levels of support for preserving the range and precision of numeric values. Developers need to be aware of the limitations of the default implementations and may need to use custom handling or configuration to ensure the desired behavior for their use cases.

[03] Summary of findings

1. What is the summary of the key findings from the article? The article concludes by summarizing the key findings:

• JSON numbers are defined syntactically in the standards, but the actual implementation behavior can vary widely across programming languages and libraries.
• Many common implementations use the built-in numeric types of the language, which can result in precision loss or errors for certain numeric values.
• Developers need to be aware of the limitations of the default implementations and may need to use custom handling or configuration to ensure the desired behavior for their use cases.