Re: [Bitcoin-development] Safe auto-updating

2013-08-07 Thread Mike Hearn
As you're Mac specific you could just use a modified Sparkle or something like that. Even if you want to use a stock Sparkle, I have some code that does threshold RSA. My intention was to use it for the Android wallet but I never found the time. I can send you a copy if you want. But it's easier an

Re: [Bitcoin-development] Safe auto-updating

2013-08-07 Thread Wendell
That multisignature/blockchain commitment idea seems really solid, Peter. Thanks very much indeed everyone, this is all very helpful. Much to research and think about. Interestingly, a thread is presently raging on liberationtech about Tor Browser Bundle, and the subject of automatic updates ha

Re: [Bitcoin-development] Safe auto-updating

2013-08-05 Thread Peter Todd
-BEGIN PGP SIGNED MESSAGE- Hash: SHA256 Gregory Maxwell had some good ideas along these lines at the san jose conference. Extending gitian with these kinds of features would be a good approach. But I think its worth thinking about attack models. A huge danger with auto-updating is that

Re: [Bitcoin-development] Safe auto-updating

2013-08-05 Thread Jim
One approach you could use would be to use bitcoin signing on a list of the build artifacts together with their SHA256 hashes. If you have a look at the MultiBit release notes you get the overall idea: https://multibit.org/releases/multibit-0.5.13/release.txt Currently these aren't machine read

Re: [Bitcoin-development] Safe auto-updating

2013-08-05 Thread Alan Reiner
Indeed. You can hardcode a "distributor" public key in the software, and client software will only trust signed data from that key. Of course, the private key for that data is not kept on the server distributing the signed checksums. Ideally it would be kept offline, and the couple-times-per-yea

Re: [Bitcoin-development] Safe auto-updating

2013-08-05 Thread Daniel F
If you want package authentication, you should at least throw in some digital signing, not just a checksum. With a compromised host, both the checksum and binaries can be changed undetectably, but if there's a signature made by a key that is not kept on the host, there's no way to fake a valid bina