Guillaume Malette

@mahlingam would you be able to compile the js file with this change and make it public somewhere? I'm very interested, unfortunately this repo doesn't seem to compile under macOS...

I managed to compile it by shedding some dependencies and using `-std=c++17` 🙂

@jnormore JASON! LONG TIME! Sorry I hadn't seen your message, wasn't checking notifications. I do have one, and though I imagine it's some months too late, here it is https://github.com/gmalette/buy-button-js/tree/line-item-properties

Any chance this will get picked up again?

> I suppose that's what `Oj.dump(value, mode: :rails)` tries to emulate? Correct! For our dataset `Oj.dump(value, mode: :rails)` produces the exact same outcome as `ERB::Util.json_escape(JSON.dump(value))`, which I believe is identical...

Looking at the performance improvements of using `gsub`, it looks like they won't be enough to bring us in line with Oj, right? If that's the case, and if you're...

> I'm almost tempted to do a more generic API where you essentially provide the escape table, but that might be too much. Funny you should mention that, while we...

I benchmarked with [this script](https://gist.github.com/gmalette/8d1c05dc5d586fff3bb5c4ceb1e81f15) (subset of our test corpus) with ruby-head. Using `JSON.dump(value, script_safe: true) + gsub` is significantly slower than `Oj` ``` ❯ ruby --version ruby 3.5.0dev (2025-03-13T13:08:01Z...

oooooh I completely missed that from the patch! Incredible. I guess apples to apples, the Oj benchmark would also need to use the binary mode, but this is something I...

I'm not sure #293 would fix it because both sources define the methods, but I'll be honest I don't fully understand how tapioca works > I think we should fix...