Re: [xml-dev] XML Namespaces 1.1

[Rick Yorgason <rick@firefang.com>]

On 24/05/2011 11:05 PM, Liam R E Quin wrote:
> On Tue, 2011-05-24 at 22:34 -0400, Rick Yorgason wrote:
>
>> * If a prefix doesn't match a declared xmlns (formerly ill-formed) it
>> shall be shorthand for including an xmlns="prefix" attribute.
>> * Namespaces should be in reversed DNS form.
>>
>> With those two changes, namespaces are no longer awkward to use, no
>> longer difficult to teach, and the cost of adoption for existing
>> libraries is approximately three lines of code.
>
> No longer awkward - for some uses, and they become 100% totally
> completely broken for other uses.  Win some, lose some.

I'm pretty familiar with how seemingly trivial changes can cause 
unforeseen consequences, so I'm being completely earnest when I ask 
this: could you give me an example of something this would actually break?

All this proposal does is take markup that's currently invalid and 
define a valid behaviour for it.

>>   So what do we have to
>> do to get a new "Namespaces in XML" standard published with these
>> minimal changes?
> (1) get consensus that the broken cases don't matter (e.g. the people
> who dereference a namespace URI to get a GRDDL file to find a Schema)

This proposal can be defined as an alias for xmlns=.  xmlns= wouldn't go 
away.  People can still use it if they want to use a namespace that 
isn't also a valid prefix name.  It's just not the recommended as the 
best practise.

> (2) change all existing XML systems to accept your new usage instead of
> raising an error. This includes...
>
> [1] buy everyone a new television
> [2] buy everyone a new VCR (TVs and VCRs use uPNP in XML to find each
> other)
> [3] replace everyone's mobile phone
> [4] replace iPods
> [5] re-engineer air traffic control systems, aircraft navigation...
> [6] replace people's washing machines, and other XML-communicating
> appliances
> [7] change all Web Services
> [8] replace all Microsoft Windows, Gnome, KDE, OS X, desktops
> [9] car components (brakes, tyres, engines,...) communicate with XML...
> [9] (many more examples)

Surely you're exaggerating.  Nobody fully implements a new W3C feature 
until it's standardized (even when a standard is based off of existing 
technology, it's almost invariably modified during the standardization 
process), so if nothing gets standardized until everybody implements it, 
then by definition there would never, ever, be any new XML features. 
You would never have XInclude, or XLink, or even XML 1.1.

I mean, we're talking about a language that has its roots in *browsers*. 
  I think there's pretty good precedent that people understand the 
concept of "If your environment doesn't support a feature yet, then 
don't use that feature yet."

I know I probably sound a bit confrontational, but I don't think you 
actually believe that it's as hopeless as you make it out to be.  I 
mean, your own namespace proposal is quite a bit more disruptive and 
ambitious than this one, so surely you still have *some* hope that 
reform is possible.

-Rick-