nodejs or/and webassembly binary ? about simdjson HOT 9 CLOSED

Yes. WebAssembly is probably out for the time being, but a native node extension is something that ought to be done.

from simdjson.

In what size of json file we can expect the benefit? if that is known, I think it will switch in using this instead? but for me, as it is working best as a compiled binary..maybe a node script that Interface it using child_process is more suitable?

from simdjson.

Say we have a working Node binding to simdjson.

In the readme, it reads:

Performance is optimized for JSON documents spanning at least a few kilobytes up to many megabytes: the performance issues with having to parse many tiny JSON documents or one truly enormous JSON document are different.

Does this mean that, this library would not be a suitable replacement for JSON.parse(jsonstring) in Node.js, where jsonstring is just a single line of JSON string?

from simdjson.

The issue is not resolved yet, why close then ?

from simdjson.

Does this mean that, this library would not be a suitable replacement for JSON.parse(jsonstring) in Node.js, where jsonstring is just a single line of JSON string?

If you are parsing a single line of JSON, then it is unlikely that the time spent parsing is significant enough to warrant optimization unless you are parsing many, many such cases. And at that point, things like function call overhead and memory allocations are likely to dominate your running time. This is potentially interesting, but it is not something that this library was optimized for.

from simdjson.

reactjs/rfcs#68 (comment)

from simdjson.

const [isOnline, setIsOnline] = add(users, isOnline(friend.id));

from simdjson.

In what size of json file we can expect the benefit?

From the README: "Performance is optimized for JSON documents spanning at least a tens kilobytes up to many megabytes". This library (simdjson) is for processing large volumes of JSON.

from simdjson.

My statement above were deliberately pessimistic: simdjson does very well even on tiny documents. At the time I wrote the above comments, I had not benchmarked simdjson on tiny documents.

from simdjson.