Daniel (dB.) Doubrovkine

Thanks for opening this issue. [Catch All Triage, attendees [1](https://github.com/dblock), [2](https://github.com/reta), [3](https://github.com/lukas-vlcek/), [4](https://github.com/rafaelma), [5](https://github.com/krisfreedain), [6](https://github.com/gargrajat), [7](https://github.com/andrross)]

[Catch All Triage - [1](https://github.com/dblock), [2](https://github.com/reta)]

We've relied on `multi_json` to swap JSON serializers between the default one and `oj`, so removing it would be backwards incompatible and definitely create problems for users. It's possible that...

> Its true that `Multijson` facilitates the usage of Oj but anybody could simply call `Oj.mimic_JSON()` and `Oj` would override all `to_json`. It could be a relic of the past....

Looking at #2483 I think we need more tests, @manueljacob care to add one to that PR that demonstrates what you describe?

What are the downsides?

What would it take to make this package compatible with llrt? PRs welcome. > believe OS is an AWS product anyway This is not the case, especially since September, OpenSearch...

@paul-uz We would like many of these things. We've done some serious surgery as part of https://github.com/opensearch-project/opensearch-js/issues/803 and would like more. Things like moving from `http` to `fetch` is likely...

> `http` to `fetch` might be one I could attempt. Sounds great. A pure replacement would be a great start, and we should then talk about something pluggable in the...

Check whether this is still the behavior on HEAD and turn it into a (failing) PR/spec? Note that your post may need a correct `content-type`, transforming what you post to...