Re: [Json] Nudging the English-language vs. formalisms discussion forward

["Manger, James" <[e-mail]>]

One more example of an existing practice for describing JSON values appears in recent draft specs from the FIDO Alliance <http://fidoalliance.org/specifications/download>. It seems to be based on Web IDL with ECMAScript bindings <http://www.w3.org/TR/WebIDL/#ecmascript-binding>.  The following example comes from fido-uaf-protocol-v1.0-rd-20140209.pdf:

dictionary Version {
  int mj; // Mandatory.
  int mn; // Mandatory.
}

dictionary OperationHeader {
  Version upv; // Mandatory.
  DOMString op; // Mandatory. Must be “Reg”, “Auth” or “Dereg”
  DOMString appID; // Mandatory. string[1..512].
  DOMString serverData; // Optional, string[1..1536]
  Extension[] exts; // Optional.
}

dictionary Extension {
  DOMString id; // Mandatory. string[1..32].
  DOMString data; // Mandatory. base64url(byte[1..8192]).
  boolean fail_if_unknown;// Mandatory.
}


I don’t particularly like it; it doesn’t feel like JSON (eg "dictionary" vs "object"); constraints are all in comments; and IDL is aimed at APIs as opposed to protocol messages. But it sort-of works: it is fairly easy to guess what the JSON has to look like.

--
James Manger