Re: Actor model implementation, seeking feedback

[Ariela Wenner]

(so sorry about the duplicated mail, I forgot to hit reply to all)

Hey Vasilij, thanks for the feedback!

0. Haha I know, I legit found out about and got very interested in the actor
model wayyyy after I got into Scheme, and was pretty surprised when I read
they're related.

1. Why both? Well, mostly laziness. JSON seemed at the time an easy way to
package the nonce outside the cyphertext without changing the design I already
had. Very kludgy, and very likely to change in the future.

2. You're right! I actually hadn't thought about that! Guess I had a hard time
not conflating "unique and non-repeatable" with random numbers... would a
timestamp be a better (but probably still not ideal) approach?

Thanks again for taking the time to look into this thing!

Cheers!

On Sat, 26 Jun 2021 20:38:03 +0200
Vasilij Schneidermann <mail@vasilij.de> wrote:

> Hello Ariela,
> 
> > ...
> 
> 0. Funny how history repeats itself with the actor model and Scheme:
> <https://en.wikipedia.org/wiki/History_of_the_Scheme_programming_language#Carl_Hewitt,_the_Actor_model,_and_the_birth_of_Scheme>
> 
> 1. Why Protobuf? Why in combination with JSON? Why not just JSON? Or
> just Protobuf, but consider it's been designed for the cases where JSON
> is too expensive to use, so not for this project I guess...
> 
> 2. Cool that you use tweetnacl for encryption, but please don't use
> random numbers for nonces, that's just wrong. Nonces are not supposed to
> be secret, random or unpredictable, but unique numbers that do not
> repeat. Random numbers do repeat eventually. If a nonce repeats, this
> allows certain cryptographical attacks to be performed on the
> corresponding ciphertexts.
> 
> Vasilij