Lists Home |
Date Index |
Pardon, Jonathan, but your request appears to be a begging task.
Surely most of us know this. XML doesn't care. The question
is, can it be made to? The answer is, given specifications that
reference each other, processors can be made to and this can become
an implementation burden and can take away that freedom to choose
the appropriate mindset however situational it may be.
Won't happen? Sure it can.
Looking at another thread. Let's say that in the currently debated task to
"subset XML", the outcome is a new version of XML per the XML-SW.
Some like it; some don't. Some want to be sure that an XML compliant
processor always supports DTDs; some want to be sure that there is a
freedom to ignore namespace specs. Others want DTDs to become legacy
only and namespaces to be reliably and always supported in the XML
Who should be allowed to decide or is a decision really necessary?
As long as datatypes can be ignored, we are where we are now and
one can dismiss Uche's article as a bit of flamebait because nothing
has changed in the specifications. Or is a change coming? As
you say, "depending on the software you use to process it".
Datatypes are a red-eye like that. What is being struggled over
here and elsewhere is the precise and unambiguous definition of
the features that can always be relied upon from an XML processor.
Because some think XML is literally XML 1.0, others think XML is
the "XML family of specifications" and another group believes
"XML is a serialized infoset", there is quite a bit of divergence
and the process of settling these seemingly obvious perma-issues
can result in a very different XML than we started with.
That is why some of us say loudly, "Don't touch the core. The
risks are too high and the benefits not widely sharable."
From: Jonathan Robie [mailto:email@example.com]
But the beauty of XML is that you are allowed to use either mindset - the
same data may be treated as well-formed XML or as strongly typed XML,
depending on the software you use to process it.