Re: [xml-dev] Is Web 2.0 the new XML?

[Robin Berjon <robin.berjon@expway.fr>]

Michael Kay wrote:
>>No. We need XML Schema so that other WGs know what not to do. 
> 
> OK, so what exactly did they do wrong, and how can other WGs avoid doing the
> same?

I don't claim to know everything that went wrong with the Schema WG, but 
here's a quick list of what springs to mind (most specs suffer from at 
least one of these, but XML Schema is special in that it has them all):


  Version One Syndrome. Since everyone wants their pet feature to be in 
the first version, which then becomes the humongous blob that we have 
come to know and loathe. Doing this, the WG never has a chance to do a 
reality check on its requirements before it's way too late to change 
anything.


  Space Odyssey Sydrome. It was a brilliant idea to separate parts 1 and 
2. Unfortunately, there's a lot more that should be separated out of 1 
(Rick posted a list of good examples recently though I can't seem to 
track down the message). A less monolithic spec would likely be easier 
to implement (certainly simpler for those of us who only need subsets) 
and perhaps simpler to fix.


  Pompous Vagueness. The spec, especially part 1, works incredibly hard 
on being thoroughly impossible to read, yet no obvious formalism appears 
when reading it. XQuery for instance learnt from this mistake.


  Usability Testing. It may sound silly but checking that the people who 
will be using your spec are actually able to do so is generally a good 
idea. I'm not the smartest monkey on the block, but generally when you 
toss me a specification in the Web and/or XML space I understand it on 
the first pass and after using it regularly for two weeks I only need to 
look up the odd detail (that's definitely the case with RelaxNG). I've 
been using XML Schema for three years, on a daily basis, and I'm still 
not familiar with it. It's guaranteed that if I hand-author even the 
simplest schema I'll make a mistake. The accumulated time I've spent 
trying to answer XML Schema questions by going through the spec with a 
thin comb is measured in months yet it's so convoluted and obfuscated 
that I'll forget the answer after a while.


  Tools will solve it. A lot of the spec seems to have been designed 
with tools in mind rather than text editors (the human-hostile syntax 
springs to mind), which is also known as the Sweep It Under The Rug 
school of design. When creating an XML vocabulary, apart from a very few 
exceptions, and your users require tools to use it, you have a problem.


There's probably more, but that's all I'm thinking of right now. I think 
it's an interesting experience to do. But then I guess I should just 
quit whining and get a job that doesn't require me to deal with that 
specific gorgon (so long as we can keep it from polluting other specs at 
random) :)

-- 
Robin Berjon
   Senior Research Scientist
   Expway, http://expway.com/