Re: Is validity the real issue?

On Nov 7, 2005, at 11:17 AM, Jens Meiert wrote:

> No. There is also some probability that you cross the city  
> unharmed, as
> there on the other side is the risk that an invalid page might fail in
> certain UAs. And this risk ain't quite low, as you need to  
> anticipate a
> future where more and more UAs behave more and more strict. So that  
> risk is
> definitely high, and the metaphor I used proves useful, I fear.

I thought about your analogy last night, and stupidly deleted the  
message I wrote in response because I should only be picking so many  
fights at one time. But there's one thing I should mention here,  
relative to your analogy:

If a government hires a company to build a road from City A to City  
B, and they build the road in a manner that somehow fails to meet the  
government's standards, then the way to resolve it is to keep the  
road closed until it meets those standards (or, if it's not a risk to  
safety, to open the road to traffic on a few lanes, while the company  
repairs what is broken).

But if the government is concerned the road is unsafe, it doesn't  
then open the road, and write a law requiring all the cars on the  
road to shut down when a driver wants to drive on it. The car  
operates for the user's benefit, not the road or the construction  
crew or the government. Cars are also designed to deal with problems,  
within certain tolerances: that is, they don't explode when you hit a  
crack in the road, or when there are 28cm of asphalt on the road  
instead of 30cm.

That's my beef with instructing user agents to shut down when they  
encounter malformed content. It punishes the user for the sins of the  
content producer, no matter how small. The best way to enforce well- 
formed content would be to have the server check all fragments that  
are published, before they are made available to users.

-
m

Received on Tuesday, 8 November 2005 01:49:31 UTC