Dimitre,
when you write the following:
"then we have achieved "world domination" -- our "infospace" covers any data.
While this seems like grandiose architecture, I personally find it too-general to be particularly useful."
then you must have a picture before your mind's eye which is quite different from mine. Mine is so practical that the usefulness is obvious. Let me try to
explain.
Imagine a bunch of directories cluttered with Java properties files - perhaps the distributed configuration of an Enterprise Service Bus. You are interested in the question which files set a particular property "script" to which values.
Let us assume there is a standard function "fn:propertiesDoc" parsing a Java properties file into a node tree. If
you have a list of all URIs (as one can create with a three-line XQuery script, using the file:list function of BaseX) - then a nice report is deliverd by this expression:
string-join( doc('catalog.xml')//uri/fn:propertiesDoc(.)/*/script/concat(root(.)/document-uri(.), ' : script=', .) , "
")
Now the interesting question is how
one would solve the problem without XML technology, for example with Java. If the alternatives are less elegant, more time-consuming and more error-prone, the usefulness of parsing non-XML into node trees has been illustrated.
No grandiose architecture, no world dominion. Only efficiency and elegance when dealing with distributed information - enabled by the info space in its broadened version, which includes Java property files (fn:propertiesDoc).
Kind regards,
Hans
