I'm finding that when Nokogiri (v1.5.5) parses the request_document it returns the names of child nodes with the namespace prefix already removed. Specifically, with this request_document (from goodreader):
The returned names of the child nodes of <D:prop> are just 'getlastmodified', 'getcontentlength' and 'resourcetype', and are therefore ignored when the following find_all operation tests for the prefix. Perhaps there are more complex cases where the prop node has other children? But in my case If I remove that test everything works very nicely, including folder sync.
Thank you for the library: it has just saved me a lot of time when I needed it most.
relax propfind name discovery a bit
adds a prepare step between authentication and request-processing