Re: [xml-dev] XML attributes are weird

[Steve Newcomb <srn@coolheads.com>]

On 12/22/2013 05:59 AM, Alain Couthures wrote:
...
> Yes, 'name' is a sort of mandatory attribute for elements. JSON doesn't
> define any mandatory property for objects.
...

JSON is fantastically useful.  I use it all the time, and more every
day, it seems.

But if I need a data/metadata distinction, JSON doesn't cut it.  XML is
the interchange language of choice for that.

The reverse is also true.  XML is not the best choice if its
data/metadata distinction stands in the way of progress.

IOW, I, like Simon, agree with Mike Kay when he says,

"For a language designed for data interchange, the attribute/element
split would be nonsense. But for a document markup language, it makes
eminent sense."


I think it *may* be useful to mention here that the HyTime term "GROVE"
(officially, "Graph Representation of Property ValuEs", ugh) was
intentionally meant to acknowledge that in the SGML parse tree (the
"grove"), each element can sprout an extra tree.

In a future, better world,

* "metadata" will stop being a synonym for "especially underprivileged
data".  IMO, there's no basis for believing that metadata doesn't
deserve metametadata, etc.

* metadata will have the privilege of rooting a robust hierarchy,
instead of the stunted one offered by XML attributes.

* no one will think that metadata are in any sense less important for
effective data interchange than the data themselves.

Viva la data/metadata distinction!

(I can't help noticing that the White House evidently thought it
wouldn't bother anybody if "just the metadata" of all the phone calls in
the world were collected in secret.  That attitude reveals something
about the widespread upside-down view of the data/metadata distinction.
 Such a view is bound to die out.  The problem is basically that we're
currently so used to the idea that metadata are underprivileged.)