RE: New issue: error recovery practices (Re: Proposed TAG Finding : Internet Media Type registration, consistency of use)

Not oddly, it is yet another risk management issue.  HTML 
in and of itself is not risky as long as the browser vendor 
manages the risk.   XML Draconian management says the risk 
accrues to the sender and that the receiver will be informed 
immediately if a sender defaults.  Then local policy takes over 
that is, one can fix the transmission locally, inform the 
sender, cut off the head of the IT department, whatever.  Over time, 
such policies might improve conditions but not immediately and 
not without costs.

Keith is right that they will ignore a specification for 
HTML that changes the owner of the risk because that means 
the owner of the cost of the risk changes as well.  If 
you ever want that to change, the owner of the cost must be 
persuaded the cost will come down by such a change.

Good specifications do not thrive based on remaining silent 
about use risks.  Namespaces is the stellar example of the 
failure of such policies.  Rather than remaining silent, 
it would have been better to informatively state that 404s 
are the result of dereferencing.  Yes, that would have 
forced a change later should the notion of dereferencing 
become attractive, but that again, is risk management. 
It might have resulted immediately in an action item 
for a policy that states clearly what behavior is to 
be expected on dereferencing the namespace.

len

-----Original Message-----
From: Chris Lilley [mailto:chris@w3.org]

On Saturday, June 1, 2002, 12:56:30 AM, Keith wrote:

KM> otoh, a requirement in the specifications to change functionality in 
KM> a way which causes more pain to users (e.g. forbidding browser 
KM> interpretation of improperly-labelled content) is highly likely 
KM> to be ignored.  

That makes it sound like an absolute. "more pain" is easy to argue, in
practice it is not a question of more or less pain, which would be
easy, but where the pain shows up and how long after the content
originator has moved on to other things (ie unmaintainable content, or
content succeptible to unexplained mysterious breakage in areas
apparently unrelated to actual changes).

Received on Monday, 3 June 2002 10:31:31 UTC