Re: ICADD support in XML [Was: SGML declaration for XML]
>From: firstname.lastname@example.org (Murray Altheim)
>"W. Eliot Kimber" <email@example.com> writes:
>>I'm not sure it's strictly necessary to include LINK in the XML spec as any
>>valid XML document can be processed by an SGML parser that supports LINK in
>>order to generate a new XML instance with the link attributes integrated.
Although that new XML instance could be provided as a special
case output of some application, an objective I believe we should have is
that any XML application should make easy the inclusion of ICADD support.
Some believe that LINK is the way to include ICADD. By that argument, W3C
removed the support for the #FIXED SDA attributes from the HTML 3.2 DTD.
Getting support for LINK presumes that some external pressure,
such as the Americans with Disabilities Act, is sufficient to
encourage vendors to supply support for the LINK feature
that is not now well-supported. Such pressure has influenced
Microsoft, and Adobe in other areas of accessibility. I hope that XML will
facilitate inclusion of ICADD capabilities.
>My experience with SP is that nsgmls requires that one modify the SGML
>declaration to include LINK IMLICIT YES in order to process LPDs. Am I
>wrong or misinterpreting your response? This sounds like a workaround
>rather than a change to the advocated specification.
That appears to be needed, as a minimum. The support that
goes behind it is also needed.
>>Note also that the "multiple attribute lists" facility used in
>>architectural meta-DTDs and almost certainly to be included in SGML97 (and,
>>I would presume, immediately adopted by XML), will provide almost the same
>>facility for adding attributes to elements unilaterally that LINK provides
>>Note that would you would *NOT* get would be the ability to apply different
>>attributes to the same element type based on hierarchy, which you can do
>>quite effectively with LINK.
I believe that the LINK capability to support hierarchy-dependent
distinctions is important.
Being able to determine where one is in context is much harder
for someone who must depend on memory, without visual cues.
Also, when one uses search to jump to a pattern match, the
context of the result is not evident, without a means to
determine that context.
>I agree that adding an LPD-like document in the declaration subset is a
>potential solution, but not as effective as an LPD, particularly as the
>HTML/ICADD LPD will get increasingly complex in attempting to resolve the
>thornier aspects of HTML 3.2, 3.5, 4.x, and I assume an XML/ICADD LPD would
>be similar in scope and complexity.
>Perhaps allowing LINK but placing specific restrictions on its use? I'm
>certainly not trying to open an enormous door here, just enough of a crack
>to provide an ICADD solution.
I believe Murray's experiments using LINK are an essential step
before any decision on how to bring ICADD up to date can be made.
At this time, we should not preclude the use of LINK from XML.