When an xml-file (the one of URL above was my example..) changes on the server to a shorter content and the URL of that file is reloaded, Then mozilla (sometimes?) appends the chars at the end of the old content after the end of the new content (to fill up to the old Content-Length). This always results in a non-parsable XML doc, as additional chars after root XML-element break the doc. Leaving the URL, flushing the disk cache and reloading works: the newly read content does not have the superficious chars at the end. But leaving URL and "Debug->Flush Memory" or flushing the memory cache does NOT result in a correct and parsable loaded XML content. Similar effect occurs when I shorten an XSL-style sheet of an XML-file: shorter XSL style sheet is broken until I leave the referencing XML-URL, flush disk cache and reload.

*** This bug has been marked as a duplicate of 79983 ***