When processing xml data 'directly' (i.e. not using the document() function) text nodes that consist of nothing but white space are missing in the output.
In most cases this is no problem because these nodes are only present due to pretty printing and can be considered as ignorable white space. However, in Trados TTX data a sequence of tab stop characters can represent 'real' tab stop characters in the source/target document and should be preserved.
There is no way to suppress this behaviour by inserting an <xsl:preserve-space> instruction, beacause an error will be thrown: "White space cannot be stripped from input documents that have already been loaded. Provide the input document as an XmlReader instead."
The funny thing is: When the same data is passed over by means of the document() function all white space is preserved; the resulting file looks exactly the same as the output of Saxon, XSLTPROC, or any other XSL processor.
