jQuery Forum

We're just using the native parser method provided by the browser and the validation RegExp from the official JSON parser - any failures are (or should be) failures in the JSON itself.

I copied the entire test page and removed the json2.js script to test this with native JSON.parse() instead of jQuery.parseJSON(). When I did this the exception is no longer thrown

The spec is here, it looks like you quoted incompletely:

http://json.org/

Any UNICODE character except " or \ or control character

Since a newline is a control character (control-J, ASCII 10 decimal), it is not allowed. Notice that the specification provides a \n escape for newlines.

So, does that mean that crockford correctly rejects this and the native JSON.parse() incorrectly accepts it?

Since Crockford defined JSON I think he was pretty sure how to parse it too.

I am not sure how someone arrives at a JSON string with literal newlines in strings. A compliant JSON.stringify implementation shouldn't produce it. That leaves some sort of ad-hoc JSON generation, which is not a good idea for so many reasons.

When the JSON string fails the regexp test, jQuery.parseJSON calls jQuery.error("Invalid JSON ...") which by default throws an error. You mentioned "silently failing" but earlier you said it throws an error (which it should). Or were you saying that the native JSON implementation silently failed on embedded newlines?

Heh, even the great Crockford is fallible, although slightly less than Chuck Noris.