Lists Home |
Date Index |
> Karl Waclawek scripsit:
> > If Unicode allows strings to have different spellings, than this is a generic
> > problem for all applications processing Unicode strings. So why add the extra
> > complexity to an XML processor to check for normalization, so that an application
> > that would normally treat Unicode strings in a standard way suddenly can do it
> > differently, because the XML processor already takes care of part of it?
> It isn't really enough to take an XML document and just check normalization
> as if it were a plain-text Unicode document. The various strings in
> XML documents -- names, runs of character content, attribute values, etc.
> need each to be checked separately.
That can be done by the application when it receives the output
from the XML processor. And even then I would assume that it is only
really necessary to check normalization when it can affect the result
of operations performed on the data.
Now, this would change if normalization would also affect
the internal behaviour of the XML processor, but I don't think
this is the case, or am I mistaken?