You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
I have searched existing issues to ensure the feature has not already been requested
🚀 Feature Proposal
fast-json-stringify does not work with TypedArray objects (Float32Array, etc.). Normally, using JSON.stringify to serialize a TypedArray results in writing it as an object (e.g., new Float32Array([0.1,0.2]) gets serialized as {"0":0.1,"1":0.2}). However, since fast-json-stringify uses a JSON Schema to guide its output, it can know that a given property is an array; it seems reasonable and consistent to allow it to serialize TypedArrays as well as regular arrays.
TypedArray can only hold (floating point or integer) numbers, so it would not work with many JSON Schemas. One option would be to only allow TypedArray if the JSON Schema is compatible with TypedArray (e.g., only number items); another option would be to allow it as is and then let individual elements fail validation.
If a TypedArray is in use, largeArrayMechanism is ignored (treated as default).
Motivation
Our application uses TypedArray for several purposes. This is fast internally but is noticeably slower to serialize. Using fast-json-stringify, locally patched to support TypedArray, is over 2.5x faster for a Float32Array of 4000 randomly generated numbers.
Prerequisites
🚀 Feature Proposal
fast-json-stringify does not work with TypedArray objects (Float32Array, etc.). Normally, using
JSON.stringify
to serialize a TypedArray results in writing it as an object (e.g.,new Float32Array([0.1,0.2])
gets serialized as{"0":0.1,"1":0.2}
). However, since fast-json-stringify uses a JSON Schema to guide its output, it can know that a given property is an array; it seems reasonable and consistent to allow it to serialize TypedArrays as well as regular arrays.Proposed implementation:
default
).Motivation
Our application uses TypedArray for several purposes. This is fast internally but is noticeably slower to serialize. Using fast-json-stringify, locally patched to support TypedArray, is over 2.5x faster for a Float32Array of 4000 randomly generated numbers.
Example
The text was updated successfully, but these errors were encountered: