Mark Pulford

Another option: Can you call `fpconv_init` after the locale has been changed? `fpconv_init` isn't thread safe, but neither is `setlocale`, so it shouldn't be any worse.

Another option.. You could use compiler defines to autodetect the correct endianness for `USE_INTERNAL_FPCONV` by adapting something like this: https://stackoverflow.com/a/58642785 It might be reasonable to add something similar to CJSON...

To help understand the impact a bit better, does anyone have examples where decodes or APIs break with: - escaped forward slash? - non-escaped forward slash? If so, which ones?...

..continuing the discussion from #36 here.. @brimworks, thanks for providing the motivating example of an API that returns larger numbers as an opaque ID field. Unfortunately it shows one of...

Erk.. wasn't thinking - numbers don't have individual metatables.. Back to: 1) Runtime option to return JSON numbers as strings for application to handle. - Pro: Numeric methods still work...

I have merged #20, does this help with your case? Are there are changes that are required to improve support other targets?

You will need to link with libm (`-lm`) for the `floor` function.

I think this would be better handled by a metatable to specify whether an object or table is needed. That would allow more consistent use of types.

This is a huge API surface without any real production testing (AIUI). Perhaps it might be better to land it under golang.org/x for some time? Eg, like context, xerrors changes.

> It's available under [golang.org/x/exp/slog](https://pkg.go.dev/golang.org/x/exp/slog) True, but it is unlikely to see much production testing under exp.