Re: [ogpx] content negotiation for LLSD over HTTP connections

[Joshua Bell <[e-mail]>]

On Jun 26, 2009, at 11:00 PM, Infinity Linden wrote:

> there are actually some OGP constructions that are single
> (non-collection) types (and thus, not RFC4627 compliant JSON.)
>
> for what it's worth... none of the browsers we tested (two versions of
> firefox, one version of safari, two versions of IE) seemed to reject
> non-collections when receiving and evaluating application/json media
> types. in other words, there seems to be a fine history in the browser
> world of ignoring RFCs.

FWIW, the ECMAScript 5 folks are formalizing that relaxation in the  
ES5 JSON parser, at least based on the latest discussions.

https://mail.mozilla.org/pipermail/es-discuss/2009-June/009481.html  
(Allen Wirfs-Brock, ES5 editor)

and:

https://mail.mozilla.org/pipermail/es-discuss/2009-June/009443.html  
(Douglas Crockford, author of the JSON RFC)

and the summary that indicates we're not crazy:

https://mail.mozilla.org/pipermail/es-discuss/2009-June/009445.html

Also in the thread, Crockford indicates no intention to issue a new  
JSON RFC. (That's my interpretation of the thread, mea culpa if I mis- 
implied.)

There is also some tooth gnashing about considering U+2028 and U+2029  
as whitespace.