1010Re: [json] MIME types
- May 5, 2008On Mon, May 5, 2008 at 7:31 AM, kriszyp <kriszyp@...> wrote:
> I am also curious if there any thoughts about JSON supersets. ThereI think one big practical problem is that there doesn't seem to be a
> have been little efforts to create any formal proposals in this area,
> constructs for JSON-esque data, and postel's law application to JSON
> flexibility (non-quoted keys, comments, etc). One could think of
way to gracefully handle the problem of auto-detecting superset in
use. That is, plain old JSON unfortunately has no place (similar to,
say, xml declaration) to add something to allow processors to figure
out if it's "plain" json or something else.
While lack of such marker makes it easier to embed json (json
encapsulation in json is trivial; in xml declaration & single-root
requirement make it very cumbersome to embed xml in xml) it's a pain
if one tries to extend markup itself.
That's why I suspect it might be best that json remained as is, and
further development was done at higher level by using what exists,
instead of trying to shoehorn new features at low level ("just because
something can be done don't mean it should").
But then again, if it was accepted that things didn't remain backwards
compatible, some features would be easy to add: things like comments
(which I miss -- there's no way for generators to add notes as to
sources, or time of generation), and have general usefulness.
Anyway: my one comment regarding supersets is that any extensions to
I really want "core" Json to remain language-agnostics (and become one
in developers' minds); and if there are extensions that are more
js-oriented, those be strict supersets which can be reliably detected,
ignore them, or downgrade processing suitably.
-+ Tatu +-
- << Previous post in topic Next post in topic >>