Re: Error handling and legacy content

Le 2 mai 2007 à 17:18, Anne van Kesteren a écrit :
> I have a hard time understanding what people actually have against  
> handling legacy content.

People rarely have something against handling legacy content. In what  
I have read they are often against perpetuating bad markup.

> What would help me a lot I think is people proposing specifically  
> what they would like and how they would solve the issues.

I think it would help a lot of people to frame the discussions with  
their implementations. This is, IMHO, the source of misunderstandings  
most of the time.

	When a feature is proposed, discussed, challenged,
	define the type of products and the conformance
	requirements for each of these products.

Consuming user agents are only one class of products. What you  
noticed is that each time a feature is discussed, it is discussed  
with the browser point of view and sometimes consumer user agents.  
Then, people perceive that the feature is discussed from their point  
of view, and there is misunderstanding.

I guess there will be a lot less heated discussions if it was better  
explained at the start.


# feature X:

## definition
   the definition
   options, etc.

## requirements for user agents consuming the code
    General requirements

    * specific visual browsers
    * specific vocal browsers
    * Search engines
    * etc

## requirements for user agents producing the code
    General requirements

    * specific for scripting engine (in-browser or not)
    * specific for visual authoring tools
    * for human authors
    * etc.


For example, a validator/conformance checker MUST be able to read the  
legacy/invalid content to help people to produce a good code. The  
same for tidy. I'm pretty sure nobody will go against that.



Another fear about legacy content for people is very natural. Not  
wait let me rephrase that.
	"anover feer aboot legassi contente four people ize very naturale."

As a human I will be mostly able to parse the bad english, but should  
it be accepted by any user agents, should it be transmitted and  
repeated over and over. Browsers have impacts on how the code is  
produced, more sometimes than they believe. (text/css for CSS, </ 
table> are some examples from the past.) Should I continue to use bad  
English and stop trying to improve because anyway, people will be  
able to parse my bad English in most cases.
	Parsing model for user agents is a good thing.
Draconian handling in some cases might be proved useful too, for  
example, in authoring tools, CMS and code producing scripting tools.


-- 
Karl Dubost - http://www.w3.org/People/karl/
W3C Conformance Manager, QA Activity Lead
   QA Weblog - http://www.w3.org/QA/
      *** Be Strict To Be Cool ***

Received on Tuesday, 8 May 2007 04:28:46 UTC