Hi Robert, *, On Tue, Oct 25, 2011 at 5:05 PM, Robert Burrell Donkin <[email protected]> wrote: > On Tue, Oct 25, 2011 at 1:38 PM, Christian Lohmaier > <[email protected]> wrote: >> On Tue, Oct 25, 2011 at 2:15 PM, Robert Burrell Donkin >> <[email protected]> wrote: >>> On Tue, Oct 25, 2011 at 12:36 PM, Christian Lohmaier >>> <[email protected]> wrote: >>>> [...] >>> Better to download the signature over HTTPS but yes, I see no reason >>> why this approach could not be made to work >> >> With signature I meant a real signature (gpg signature), not a md5sum >> or sha1sum file. >> When it is a cryptographic signature, it doesn't matter how you >> download it, as it cannot be faked. >> (of course the user has to get the proper key, but that's a different issue) > > FWIW it's a defense in depth measure[1] >[...] > [1] Consider an attacker with some ability to fabricate convincing > signatures.
Define "convincing signatures". If anyone were to be able to create convincing gpg singatures of Apache releases, then this... > Downloading the signature from a trusted server means that > such an attacker would need to replace an existing signature on secure > hardware without detection. is moot anyway, the lesser problem to be concerned about. And this btw. is not any different than to download the torrent via https. >> So it is not a matter of infrastructure, but a matter of policy. > > Where's the URL for this policy? I didn't mean to imply there was a set-in-stone policy already. What I meant was that it is up to the project to decide whether torrents are used or not, that the technical implementation of using torrents is so simple that apache infrastructure is not needed at all. You want torrents, you got torrents. You don't want them, you just don't use them. (Of course I don't know whether Apache as a whole has a written policy or guidelines wrt. using torrents, but I don't think there is one) ciao Christian
