Re: [xml-dev] The year is 2027, and we need to examine archived X ML doc

["Brendan Macmillan" <melbourne.research@gmail.com>]

To see what might replace XML, we could consider what made XML a success.

As a tree-based data format, there's nothing special about XML - except perhaps that it can markup documents in a natural way (ie. text as opposed to data). Many alternative technologies could have won instead. I agree the advantages of XML (or something like it) are enabled by cheap cycles and memory:
  XML is pre-parsed (in contrast with something like parsing C source code).
  XML has meta-data (a step towards declarative semantics).

I think XML became popular because HTML became popular ("What's this XML thing?" "It's HTML with your own tags"), and HTML become popular because of the web. XML just rode the tidal wave (not criticising XML - it's just how technologies get adopted).

But today, I can't imagine XML being displaced. It's like ASCII - the question isn't "Why would we use ASCII?", but "Why wouldn't we?". There once were competing alternatives to ASCII, but not today. Instead, we have Unicode, which subsumes ASCII, instead of displacing it.

XML is the new ASCII. 

(I think) whatever comes next will be built on top of XML, just as XML is built on top of Unicode/ASCII. It will meet a need that XML does not meet very well; and will be facilitated by advances in technology that were previously unthinkable; but this new thing won't actually be adopted until there is a new tidal wave to carry it. I have no idea what that wave will be - maybe direct brain integration, like cochlear implants. Who knows?

As far as the particular XML-subsuming technology goes that is carried by this wave, it could be do with the needs of integration and interoperation. Specifically, it could be declarative semantics. The big problem is that semantics are very hard to capture in the first place - let alone represent them. The beginnings of a solution could come specific vertical industries, as their processes become better understood and standardized. Looking at the these experiences might lead to a general solution, of how to represent the semantics that are typically needed in practice in a way that is theoretically consistent. 

So, my (wildest) guess is:
Declarative semantics subsume XML.


cheers,
Brendan

On 12/09/2007, Cox, Bruce < Bruce.Cox@uspto.gov> wrote:

I'd think that something higher in the stack would be more likely to
supersede XML.