Closed baryon closed 4 months ago
https://gist.github.com/baryon/6a0f97fc55a39b826d6e3d40166afa59
I noticed that bigint can be interpreted as int64. I have another proposal, what does everyone think? That is to use the hex of bigint, and use a byte array to represent bigint. Of course, we also need to consider negative numbers. I wrote code and test code to perform the conversion between bigint and buffer. Is there anyone who can review and integrate it into this codebase? I'm not quite clear on which format should be used in msgpack to represent it.
https://no2chem.github.io/bigint-buffer/
a better lib than my code, it support Big/Little ending
Add bigint
private encodeObject(object: unknown, depth: number) {
// try to encode objects with custom codec first of non-primitives
const ext = this.extensionCodec.tryToEncode(object, this.context);
if (ext != null) {
this.encodeExtension(ext);
} else if (Array.isArray(object)) {
this.encodeArray(object, depth);
} else if (ArrayBuffer.isView(object)) {
this.encodeBinary(object);
} else if (typeof object === "object") {
this.encodeMap(object as Record<string, unknown>, depth);
} else if(typeof object === "bigint"){
this.encodeBigInt64(object);
} else {
// symbol, function and other special object come here unless extensionCodec handles them.
throw new Error(`Unrecognized object: ${Object.prototype.toString.apply(object)}`);
}
}
I am using useBigInt64 flag.
const encodedData = msgpack.encode(data, {
sortKeys: true,
useBigInt64: true,
});
When I set a number is 2100n 10000n 10n ** 8n, got an error