The problem is that the mime types of files that are to be downloaded are not detected correctly.

E.g., a pdf file can be downloaded with the "save" command, but the file name gets the extension "html".
Also the "shell-command-on-file" works, you may type in a pdf viewer manually. Hence it seems that when
following a link to a pdf file, conkeror assumes it is a html page, tries to display it as a webpage, and
crashes.

Anybody knows how conkeror determines the mime types? Maybe some external program or service is used for
this purpose?