-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA256
That is great.
One thing which could help identifying the bugs faster: Tor gives too much importance to ed25519_master_id_public_key (if it doesn't see this file, it will generate a new unrequested identity regardless if it has a valid signing_cert and signing_secret_key and/or even if it has the ed25519_master_id_secret_key unencrypted).
These commands would be useful as well: --getpubkey; --encryptkey; - --decryptkey; --newpass; --gensignkey.
On 8/6/2015 4:14 AM, Nick Mathewson wrote:
On Tue, Aug 4, 2015 at 8:24 PM, s7r s7r@sky-ip.org wrote:
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA256
On 8/4/2015 5:42 PM, Nick Mathewson wrote:
Hi, s7r!
This is an impressive writeup; thanks!
One thing that makes it hard for me to follow this document is that I'm not sure which parts are describing how things work _now_, and which parts describe how things _should_ work. Would it be possible to split up descriptions of current/revised behavior, and mark each?
many thanks,
Hi Nick,
My pleasure. Done and written. It looks long but I hope it reads fast and makes sense. I know it's a PITA to fix things by reading about them on email, so do let me know what are the priorities and how should I open tickets on trac to be cleaner and easier for you (a single big ticket? more tickets for each case to the master ticket where we discuss the Ed25519 FAQ?).
Thanks; this is incredibly helpful!
I've started a branch to do a test case to demonstrate all these bugs ; it's called "ed25519_keygen" in my public repository. It also adds a couple more features to '--keygen'. It does cases 2...4 so far; I want to make it cover 5..10.
Once I've got it demonstrating all the cases, I'll try to clean it up and start hammering down the bugs.
best wishes,