It is the remote document server's responsibility to interpret the URL it receives when the cross-reference is followed and, typically, to return the correct URL (or the document itself) to the browser. The server will therefore have to search for the document in whatever form it takes on the remote machine (it may not necessarily be in HTML format). This may involve similar steps to those that the HTX linker would have performed had the document been available locally (e.g. looking the reference up in an index file), except that they are performed at run-time rather than in advance when the document is linked. The server should also issue any necessary diagnostics -- about invalid labels or documents that cannot be found, for example.