Re: p:load?

>DTD validation is intimately associated with parsing. A p:validate-dtd step
would, if I understand what you're suggesting, accept as its input an XML
document (infoset, object model, whatever the >implementation uses) and by
the time you've built one of them, it's way too late to do DTD validation.

 

I don't get it. If the input XML document doesn't already have an associated
DTD that was also parsed at parsing time, how can it be too late to do DTD
validation? And if it does, what's the problem of validating it against a
second one (other than the performance loss of course)? I mean, this step
should allow a validation with a specified DTD, rather than the one
associated with the document. A clever implementation would automatically
finish the step (with no validation) if the supplied DTD is the same as the
associated one (provided the associated one has been validated against at
parsing time).

 

Regards,

Vasil Rangelov