NEM (XEM) Official Thread - 100% New Code - Easy To Use APIs

[jabo38]

The NEM devs aren't messing around 

[makoto1337]

elliptic curve: "Switched elliptic curve from secp256k1 to ed25519: Ed25519 is a state of the art elliptic curve. Verification of signatures is about 6 times faster for ed25519. This will speed up the synchronization process."

Last time I looked it up, there were no good (reviewed) / fast implementations of Ed25519 in Java. What library are using for Ed25519 ?

That is correct. Unfortunately there were no good java implementations of this curve so we had to do it ourselves.

[jabo38]

The NEM devs aren't messing around  

it's real time

[Eadeqa]

elliptic curve: "Switched elliptic curve from secp256k1 to ed25519: Ed25519 is a state of the art elliptic curve. Verification of signatures is about 6 times faster for ed25519. This will speed up the synchronization process."

Last time I looked it up, there were no good (reviewed) / fast implementations of Ed25519 in Java. What library are using for Ed25519 ?

That is correct. Unfortunately there were no good java implementations of this curve so we had to do it ourselves.

How do you know you implemented it correctly?  Should it not be reviewed by some crypto guru who understands ECC?

[jabo38]

NEM has 5 very well educated and very experienced core devs.  There might very well be a guru or two on the team.  Apparently building a new code from scratch and doing it well is quite a challenge.

As pretty much everyone knows, I don't know hardly anything from the technical perspective, but I talked to a programmer who independently decompiled the code and was very impressed.  

Ethereum had a massive pre-sale and so it is open source pre-launch, but pretty much only the coins with massive pre-sales are open-source pre-launch.  

That said, it is not completely locked away either.  It is possible to decompile it and have a good look.  @Eadeqa, please take a moment to decompile it and take a look.  I hope you can look at it and come back and say good things!

[Eadeqa]

NEM has 5 very well educated and very experienced core devs.  There might very well be a guru or two on the team.  

There is no cryptographer among the 5. They are all programmers.

@Eadeqa, please take a moment to decompile it and take a look

I pretty sure they implemented (probably just copied C++ version) of ed25519 correctly, but still are they 100% sure?

[makoto1337]

Any Greek NEMsters?

[BloodyRookie]

For those of you who are asking themselves what the starting point for the ed25519 implementation was:

http://lmgtfy.com/?q=Bloody+Rookie+ed25519

[gimre]

For those of you who are asking themselves what the starting point for the ed25519 implementation was:

http://lmgtfy.com/?q=Bloody+Rookie+ed25519

What you'll find there is only part of our changes, it'll all be visible, once nem.core will become public
(so that NXT will be able to borrow ed25519 if they'll want to)

There is no cryptographer among the 5. They are all programmers.

We went thorugh curve25519 paper, and ed25519 paper (and bloodyrookie went through some more papers). It's math. We're NOT BAD at math.

[Eadeqa]

For those of you who are asking themselves what the starting point for the ed25519 implementation was:

http://lmgtfy.com/?q=Bloody+Rookie+ed25519

I saw that before but wasn't clear this version is being used in nem. You yourself wrote:

https://github.com/BloodyRookie/ed25519-java

"There are no guarantees that this is secure for use. Tests against the data from the Python implementation are passing, but this has not yet been audited by a professional cryptographer."

[gimre]

http://lmgtfy.com/?q=Bloody+Rookie+ed25519

"There are no guarantees that this is secure for use. Tests against the data from the Python implementation are passing, but this has not yet been audited by a professional cryptographer."

You do realize, this is text from FORKed implementation...

[Eadeqa]

(so that NXT will be able to borrow ed25519 if they'll want to)

Pretty unlikely Nxt would switch to anything else, at this point. It's probably impossible now.

We went through curve25519 paper, and ed25519 paper (and bloodyrookie went through some more papers). It's math. We're NOT BAD at math.

The basic rule is that you don't implement your own crypto and use well tested implementation.

[gimre]

The basic rule is that you don't implement your own crypto and use well tested implementation.

Do not mix implementing OWN crypto, with porting implementation.

[BloodyRookie]

@Eadeqa: it is a port from libsodium ref 10 implementation: https://github.com/jedisct1/libsodium/tree/master/src/libsodium/crypto_sign/ed25519/ref10

[pandher]

NCC working, NIS not available?

[jabo38]

Someday all these names on this page are going to be either a "hero" or a "legend". hahaha

[BloodyRookie]

NCC working, NIS not available?

In case you just started the software, NIS needs some time for startup. Have patience

[rockethead]

Why reinvent the wheel and implement one's own crypto when there are so many out there? Any crypto implemented their own? I think what our rocket scientists did was the right path. I'd do the same if i were to design a crypto.

Edit: From my deduction and observation, I believe BR was/is a professor in mathematics or physics, or at least some known and important academic figure in that discipline.

[jabo38]

While I have the devs attention.  I would like to make a request with multi-sig.  I know you guys like tackling the impossible so here is a challenge (but maybe wait until after launch, right   )

1. In Bitcoin and other coins, multi-sig as far as I know is just an option with in one person's wallet software and so each wallet company develops there own solution.  But like a mutlisig account started with one wallet software, can't really be used on another companies wallet's software as far as I know.  

Request 1: I think it would be really cool if our multisig was a cross third party platform solution and if it was I can think of lots and lots of neat things it could be used for and ways it can open up the blockchain.  For instance, smart contracts could become sooooo much easier.  I could make a "If this then that" parameter machine that could examine millions of different criteria and options and would release a part of the mutisig if some kind of function happened.  

2. All this business with multisig and multikeys is complicated for the user.  What I would think is really cool if one of the mutlisig's options could be a Google authentication code.  

Is there anyway to slip the Google Authenticator code into multi-sig? Or am I just asking too much?  (please note the do not delay launch request   hahaha)

[pandher]

NCC working, NIS not available?

In case you just started the software, NIS needs some time for startup. Have patience

Working Now