Conversation

Notices

1. Embed this notice
   Soatok Dreamseeker (soatok@furry.engineer)'s status on Thursday, 30-Apr-2026 00:17:55 JST Soatok Dreamseeker

   • Elias Mårtenson
   • Chee Aun 🤔

   @eltrac @loke @cheeaun What does faking commit authorship actually buy an attacker?

   • Embed this notice
     Soatok Dreamseeker (soatok@furry.engineer)'s status on Thursday, 30-Apr-2026 00:27:01 JST Soatok Dreamseeker

     in reply to

     • Elias Mårtenson
     • Chee Aun 🤔

     @loke @eltrac @cheeaun Your suspicion is incorrect. I'm trying to make sure I understand the shape of the problem you're trying to solve.

     You're tying commit authenticity to long-lived keys. You're stapling non-repudiation into your designs. You're assuming a lot of security properties about the cryptography algorithms involved.

     For example: Signatures do not, by themselves, provide a property called exclusive ownership. https://www.bolet.org/~pornin/2005-acns-pornin+stern.pdf

     Many of these properties matter significantly in some use cases. I was probing to see if they mattered here.

   • Embed this notice
     Elias Mårtenson (loke@functional.cafe)'s status on Thursday, 30-Apr-2026 00:27:02 JST Elias Mårtenson

     in reply to

     • Chee Aun 🤔

     @soatok @eltrac @cheeaun it's reasonably easy to think of various scenarios. But I suspect that whatever I say, the reply will be "that's not a real problem" or "there are other ways to protect against that".

     Suffice it to say that I'm not worried about specific scenarios that I can envision. It's the ones that may happen in the future I'm concerned about. Proving the chain of changes is what I think itbis is just a good idea. Especially since my primary project uses SHA1 hashes.

   • Embed this notice
     Soatok Dreamseeker (soatok@furry.engineer)'s status on Thursday, 30-Apr-2026 00:31:13 JST Soatok Dreamseeker

     in reply to

     • Elias Mårtenson
     • Chee Aun 🤔

     @loke @eltrac @cheeaun If you're using SSH signatures with Ed25519 (very common), you have Universal Exclusive Ownership by design.

     If you're using anything else, it's a "maybe". Most PGP keys I've seen, historically, were RSA keys.