Re:IOTA: A tangled mess
by dna_polymerase on Sunday December 17, 2017 @07:39AM
(#15945517)
Ethereum has a Go implementation. The Java argument is BS obivously. The protocol needs to be valid, not a single implementation.
[ Parent
| on HN ]
Single implementation is important for consensus systems. Look at e.g. the spurious dragon hardfork.
The problem with multiple implementations is that if there is a tiny difference in behavior between them, you can only find out after damage has been done to the network via a split/fork.
You can have the exact same problem with just different versions or even configurations.
I think the idea is that the protocol is resilient to differences in implementation. A bad actor, could for all intents and purposes, manipulate the implementation to their advantage. An implementation, reference or not, is merely a convenience for users who do not want to write their own clients (which is essential for adoption).
In fact, last year an attacker launched a denial of service on the main Ethereum implementation, and the network barely hiccuped because people just switched over to one of the others.
To be fair, Ethereum also has a C++ and a Rust implementation.
Re:IOTA: A tangled mess
by dna_polymerase on Sunday December 17, 2017 @06:44PM
(#15948333)
Sure, just wanted to make it clear that not all cryptos completely rely on C++.