Re: [xml-dev] Whither XML ?

[Michael Kay <mike@saxonica.com>]

> And now as I predicted, all that complexity and incoherence has come 
> back to bite the committees, and it's their turn to complain.  Frankly 
> a junta is how I perceived some of the W3C groups, and so it's 
> interesting seeing that word turned against browser vendors.

A number of people, knowing of the HTML5 battle between the 
standards-makers and the browser-makers, have interpreted my blog post 
as a salvo in that battle on behalf of the standards-makers. It was not 
intended that way (far from it). Rather, it expressed frustration that 
the standards community has missed the point by trying to define 
client-side standards such as SVG, XForms, and indeed XSLT 2.0 that can 
only be delivered if they were tightly integrated into the browser, 
something that is unlikely to happen rapidly if at all given the 
dynamics of the browser industry. It was a mistake to assume that one 
could define complex standards and lob them across the wall expecting 
vendors to implement them at their own cost, in software products from 
which no-one earns any revenue.

I think the user community wants the facilities in these standards - or 
at least, there is a substantial market for such functionality - and I'm 
frustrated by the inability to deliver it. But I'm not asking the 
browser-makers to deliver it. Frankly, if we expect browsers to be free, 
then I don't see how we can make such demands. What I am asking for is 
for the browser to become a much more open platform, in which the 
browser-maker delivers interoperable extensibility and the rest of the 
community has the ability to decide what gets offered (and what gets 
used) on top of the basic platform: an architecture in which both the 
document markup and the functionality associated with the markup are 
open and extensible - and in which they are all extensible using 
standardized interfaces.

I've heard suggestions that this can't be done for security reasons. I 
don't believe that; if programmability can be offered through 
Javascript, then I believe it can also be offered in a way that is 
programming-language independent. It just needs a virtual machine inside 
a sand-box - preferably not a rigid sand-box, but some kind of security 
architecture whereby the access of an application to resources on the 
user machine is firmly under the user's control.

I've also heard suggestions that we can write anything we like in 
Javascript. Perhaps we can - but is it that really a reasonable way 
forward? When did we last have a computing platform that required all 
applications to be written in the same high-level language?

I don't think anyone can dispute that the browser has become a 
bottleneck in terms of moving the technology forward. We're all 
constrained to move forward at the pace of the slowest browser. Some of 
the standards eventually make it - XSLT 1.0, CSS3, and SVG are examples 
- but it takes ten years. There must be a better way.

Michael Kay
Saxonica