Justify use of charset parameter in JSON payloads about api-standards HOT 2 CLOSED

It seems this came out of WhiteHouse/api-standards#22, but there wasn't much discussion on the topic.

I was the author of that piece of our document, and it came from my experience building the Sunlight Congress API. Being able to easily and correctly view JSON in-browser was a priority for that API, and non-ASCII characters would render incorrectly in-browser without the charset parameter.

I respect Armin's opinion and keeping complexity low is always a good goal, but unless there are actual interoperability problems with a utf-8 charset, it doesn't outweigh the practical benefit to me.

from api-standards.

Cool, that seems reasonable to me!

I believe that Armin's main criticism of the charset parameter is that it can be used to encode JSON into charsets that it wasn't originally intended to be encoded into, like latin1, which basically means that clients that do purely follow the spec (and therefore which don't look at the charset parameter) would get confused.

However, because your recommendation specifically recommends passing charset=UTF-8 and UTF-8 is an encoding that spec-purist clients expect, I don't believe it will result in any interoperability problems.

Anyhow, thanks for the explanation, it makes a lot of sense. I'm going to close the issue now!

from api-standards.