Cryptography Implementation Discussion

[Aide33]

This seems really awesome so far, I wish I could help but I don't know much about crypto

[Ad Bot]

[Ipquarx]

Also, should I even bother benchmarking Red Guy's library? It's missing a lot of stuff.

I actually made mine so that Red_Guy could replace his (which was very messy and pretty slow) with something better, and he benchmarked mine and said mine was faster. So I don't think you have to.

[Lugnut]

Lug, do I need a github account to contribute to the repo?

technically, you can use one of it's authentication partner thingies

so like, you could use your google account for github.

but yes, you do need to use github, sorry.
i would, like, homehost the repository if my internet didn't suck

[Greek2me]

So apparently I had an account already... https://github.com/Greek2me


Here's the final benchmarks.

http://pastebin.com/aE5k1rY1

Ip's changes saved a tiny amount of time, I guess. It's still missing a division function. Is this a problem, or does it support multiplying by decimals?

[Ipquarx]

Haha no the actual changes I made won't show their effect until you get around to multiplying 40+ digit numbers. And no it doesn't have division, because I could never figure out how to get division for arbitrary numbers, IE not using mFloor(%a / %b) since that can easily cause errors for large numbers.

[Ipquarx]

Okay, I've been working hard and I'm just about finished division of arbitrary numbers to arbitrary precision. It's a bit slow at the moment but I'm going to be optimizing it quite a bit, but not today. (To be more specific you can now do all the operations (division is untested) with decimals of any size now.)

Lug, can you add me as a contributor to the repo? https://github.com/Ipquarx

[Lugnut]

Added both of you. I think this means you can opt to just clone Tungul/Support_Cryptography instead of the forking bullstuff.

I dunno though

[Ipquarx]

Yeah that's exactly what it means, I just pushed my changes directly to the repo.

[Lugnut]

Nice. I think that's used in the point multiplication thing, ironically.

[Greek2me]

So what are we doing, elliptic curve AES?

[Lugnut]

Elliptic curves first, from there we can branch out into whatever we feel like... one time pads generated from a shared seed, aes that uses strings of 1 and 0 for input and output, etc.

[$trinick]

There's zero point to Cryptography in Blockland. Anyone with a quarter of a brain could run any encrypted strings through the decryption function themself unless you did something with public and private key pairs and at that point you're going beyond Torque's computational limits.

You may as well just write it in PHP, crack open a TCP Object and connect -> send the string to be en/decrypted.

[Lugnut]

Did you read literally anything that was posted?

[Lugnut]

Anyone with a quarter of a brain could run any encrypted strings through the decryption function themself.

this is the point of symmetric crypto, yes.

Sending the string remotely totally defeats the idea

[$trinick]

Yes, I read the entire thread. You're not going to get any key pair encryption running on Blockland, and even if you do I could probably brute force it in a matter of seconds using C++ anyway.

Sending it to a remote server does not defeat your idea at all. It reduces the time taken to produce a high level encryption by a huge margin and completely offloads the processing for the encryption.