Re: Tests that depend on node being inserted for default attributes

I sent a message earlier which may need to be qualified (it's on the web 
archives but hasn't been forwarded yet):

The DOM WG has on several occasions discussed this issue, and it is one 
which goes back to XML 1.0, as Ray indicates.

My personal reading, not necessarily that of the DOM WG, is that default 
attributes should be present with the value specified in the internal 
subset, if the document has one [1].

For our purposes, I think we should come up with some use cases for 
having internal subsets, as Ray points out.

/Dimitris

[1] http://www.w3.org/TR/1998/REC-xml-19980210#proc-types


On Friday, March 8, 2002, at 09:05 , Ray Whitmer wrote:

> Here is my take on it, speaking personally, not for the WG:
>
> Pro: The XML specification requires processing of attributes in the 
> internal subset in some cases (see the XML spec for which cases are 
> required and which are not).  Also, since Mozilla relies on a quality 
> parser implementation, I would fully expect Mozilla to pass (or in the 
> unlikely case it did not, to be fixed promptly).
>
> Con: This is not a common practice (find me a use case on the web), and 
> there are a few complications you have to be aware of when designing 
> the test case, so whether you really want to construct the test seems 
> to be a matter of priority.
>
> Ray Whitmer
> rayw@netscape.com
>
> Curt Arnold wrote:
>
> Mary Brady wrote:
>
> [mb] Have we asked the working group what should happen if the iterms 
> were
> represented as an internal subset?  I'd rather get a concrete answer and
>
> fix
>
> the
> problem than provide a work-around.
>
>
> I'd love to get a concrete, consise statement from the WG on this 
> matter,
> too.  I have raised it on this list, but I don't know if Dimitriadis has
> either taken it back to the WG or if he can clarify the existing 
> position
> from member confidential material.
>
> Might finally be time to fire up the bug tracker from our SF alter-ego 
> to
> keep track of open issues.
>
> If default attribute nodes should be provided when for default 
> attributes in
> the internal subset, I would still recommend adding new tests for that
> behavior and leaving the existing tests as is.  Basically, creating an
> is_staff.xml (for internal subset) and is_ versions of the 9 tests.
>
>
>

Received on Friday, 8 March 2002 18:07:27 UTC