Error with Message initialization

[mobileapps23]

Hello!
Your lib perfectly matches to my goals, but I have trouble with Message from your example:
let message = Message(hex: "243F6A8885A308D313198A2E03707344A4093822299F31D0082EFA98EC4E6C89" An error occurred： Cannot invoke initializer for type 'Message' with an argument list of type '(hex: String)'

How could I init Message with some String?

[Sajjon]

@mobileapps23 So sorry for slow response, saw this just now... But I think I talked to you on StackOverflow?

Hey what kind of data do you wanna sign/verify?

There are five different initializers of Message. You ought to find a suitable one. Remember that the data should be hashed before signing. Which is why I've added convenience inits such:

init<H>(unhashed: Data, hashFunction: H) where H: HashFunction

    init?<H>(
        unhashed: String,
        encoding: String.Encoding = .default,
        hashFunction: H
    ) where H: HashFunction

and

init?<H>(hashedHex: HexString, hashedBy hashFunction: H) where H: HashFunction

[narayanprusty]

@Sajjon still same issue. Can you provide code example?

[Sajjon]

@narayanprusty look at the unit tests, e.g.: https://github.com/Sajjon/EllipticCurveKit/blob/master/Tests/EllipticCurveKitTests/CurveTests/ECDSATests.swift

which also defines convenience inits:

// MARK: Message init
extension Message {
    init(message: String) {
        self.init(unhashed: message, encoding: .default, hashFunction: SHA256())!
    }

    init(hex: String) {
        self.init(hashedHex: hex, hashedBy: SHA256())!
    }
}

[luis7lucho]

Hi, quick one, is there a way to sign without hashing the message?
I am trying to replicate org.web3j.crypto framework which has the option to sign without message hashing (and you also hash with SHA3 if you wanted to) but I can't find a way to sign without message hashing here.
Thanks

[Sajjon]

@luis7lucho Sounds like you don't really know what you are doing? We ALWAYS hash stuff in crypto. Always. You cannot sign a text document being a couple of hundreds of KB or even MB. That makes no sense. We ALWAYS hash the thing we wanna sign.

So either you already have a hash, then you can you the init in init:hashedHex:hashedBy where whatever you pass into hashedBy is ignored. I'm requiring you to pass a HashFunction (via hashedBy) to force you into thinking about how the data you pass is hashed, explicitly highlighting that everything always should be hashed.

You can also use init:rawData, which is will be the exact same thing as init:hashedHex:hashedBy. But I disencourage that.

So sounds like you don't really know what you are signing...? Which really is what you should sort ought first. What are you signing? When you know what you are signing, why not just use the initializers I encourage you to use, e.g. init:unhashedData:hashFunction which does the hashing for you and where you specify which hash funtion to use.

Makes sense?

[luis7lucho]

Hi Sajjon, thanks for the reply.
I was trying to replicate a code from the web3j.crypto library from Android wich lets you create a signature without hashing the message (it has a Bool parameter which you can pass false if you do not wish to hash the message). I am not arguing wether you SHOULD or SHOULD NOT, I agree you always should, but for this particuar scenario I needed that. I got around it eventually anyway, but thanks for your reply :)

I have noticed something on your latests commits, the test and the method to generate a signature takes quite a bit of time, a bit when generating the public key but mostly generating the signature, I have put this on a background thread but wanted to know if there is a way around that?
Thanks

[Sajjon]

@luis7lucho I will be closing this issue then, since neither @mobileapps23 nor @narayanprusty have replied.

@luis7lucho regarding performance, since the project is now a SPM package, the tests are no longer run with optimization flag. I have not found a way of using optimization with tests yet, I created a this StackOverflow question, but no reply yet - I'm pretty sure it is not possible yet... Hopefully soon!

[Sajjon]

@luis7lucho please upvote the SO question, so maybe someone posts an answer, thx!