Re: PaceMustBeWellFormed

[Svgdeveloper@xxxxxxx]

In a message dated 7/15/2004 12:11:53 PM GMT Daylight Time, rubys@xxxxxxxxxxxxxxxx writes:

> Back to the example above.  While the current specs may say it is 
> well-formed, and even proscribe how it should be interpreted, I see a 
> document that is seriously confused.  And I see the relevant draconian 
> provisions of the XML specification tossing other documents which have 
> committed significantly lesser infractions involving so-called smart 
> quotes out on their ear.
>
> So, given that documents such as the above do not tend to appear in 
> practice (can you give an example?), and have no reason to appear in 
> practice (any transcoders that understand application/xml had better 
> understand xml prologs), and that the likelihood that the current 
> precedence rules will produce the intended result is low (given the way 
> that modern web servers are configured, if the above data were served 
> statically, I would tend to trust the document prolog more than the HTTP 
> headers in this particular circumstance)... given all this, what is my 
> recommendation?
>
> My recommendation is that the presence of conflicting charset/encoding 
> information in documents served as application/xml be treated as a 
> well-formedness error.


It seems to me to risk introducing an unnecessary source of confusion re terminology.

I would suggest that Atom apply, as relevant, the term well-formedness as specified in XML 1.0 (nth Edition) and use another term for situations not covered by the XML 1.0 definition of well-formedness.

Andrew Watt