I split & merge some pdf files like this
pdftk A=content.pdf B=frontside.pdf cat B1 A5-2 output output.pdf 2>&1
The result is fine but has no XMP data anymore. Is there a way to tell pdftk from which file it shall take the XMP dataset when merging?
Since this seems to be impossible, I ended up with adding Metadata with exiftool like this
exiftool -b -TagsFromFile frontside.pdf -all:all output.pdf
for those who came here searching. But beware, the Details of the XMP-Structure will be changed, custom (non-dc) fields get lost.