User-Agent: Mozilla/5.0 (Windows; U; Windows NT 5.0; en-US; rv:1.8.0.9) Gecko/20061206 Firefox/1.5.0.9 Build Identifier: When I download an attachment to a bug, I get a file called something like this: attachment.cgi?id=78885&action=view It would be more useful to have the original filename it was uploaded as. (I'm having this problem with Gnome's bugzilla, originally filed here: http://bugzilla.gnome.org/show_bug.cgi?id=389720) Reproducible: Always

I never had such problems. Probably a local issue with GNOME's bugzilla.

OK, to be more exact, you have to view the attachment first, then right click to save the attachment. It will have the correct name.

Comment on the Gnome bugzilla: Please report this upstream (bugzilla.mozilla.org, Bugzilla product). It will actually work if you first click, and then do a save as (e.g. for patches). However, the browser does not use the servers filename if you do a 'save link as'. Usually this is solved by faking the URL with something like: attachment.cgi?id=78885/the.attachment.name.patch But that is for upstream to fix (IIRC this depends on an Apache feature being enabled).

The problem is due to using Save Link As and Firefox 1.5+ (bug 299372), wget, or curl. I think we should investigate adding the filename after the link. E.g. attachment.cgi?id=12345/some_file_name.ext. But the filename should still be 'command line compatible'. E.g. only allow a safe subset. This for easy copy/pasting in a terminal.

(In reply to comment #4) > or curl. I think we should investigate adding the filename after the link. E.g. > attachment.cgi?id=12345/some_file_name.ext. No, this doesn't work. Now the ID is "12345/some_file_name.ext", which is invalid. And we are not going to force admins to enable some feature in Apache to have it running correctly, probably because there could be some security implications. And we have no idea whether this is supported by all web servers anyway.