A webapp I've been developing allows users to upload and download a type of file which is meant to be treated as an opaque blob. My app serves it up with a file extension not commonly used for any other purpose, as well as specifying that its MIME Content-Type is application/octet-stream.
Internally, the file is a simple Zip archive containing a single compressed file. What I've found is that IE6 apparently inspects the content of the file, determines that it's a Zip archive, and "helpfully" saves it with an additional ".zip" extension. Unbelievable!
As I mentioned, this file is meant to be opaque, and we don't want users to be poking around inside the file--not because it's dangerous or contains sensitive information or anything, we just don't want to confuse them. I suggested prepending the Zip content with a magic number to prevent IE6 from recognizing it, but my manager says he'd prefer it if the file content could remain unchanged, so that knowledgeable people can rename the file and examine its contents as a zip archive, if necessary.
Is there any way to tell IE6 to keep its mitts off of the file? Or any alternative approach at all? (Alas, just not supporting IE6 at all is not an option.)
Incidentally, IE7 respects the file's name, but still identifies it as a Zip archive in the download dialog. That's better than IE6, but still less than ideal.