# Ethermint 2.0: Cosmos SDK + EVM by Chris Goes (Devcon4)

- Channel: [Devcon](https://streameth.org/devcon)
- Date: 2019-10-09
- Duration: 27:05
- Topics: blockchain, smart contracts, distributed ledger, Science & Technology
- Watch: https://streameth.org/watch/yt-VCLbS1Oks8A
- YouTube: https://www.youtube.com/watch?v=VCLbS1Oks8A

## Description

Visit the https://archive.devcon.org/ to gain access to the entire library of Devcon talks with the ease of filtering, playlists, personalized suggestions, decentralized access on IPFS and more.
https://archive.devcon.org/archive/watch/4/ethermint-20-cosmos-sdk-evm

Chris Goes presents his talk on Ethermint 2.0: Cosmos SDK + EVM.

Speaker(s): Chris Goes
Skill level: Expert
Track: Developer Infrastructure
Keywords: tendermint, dapps, architecture

Follow us: https://twitter.com/efdevcon, https://twitter.com/ethereum
Learn more about devcon: https://www.devcon.org/
Learn more about ethereum: https://ethereum.org/ 

Devcon is the Ethereum conference for developers, researchers, thinkers, and makers. 
Devcon 4 was held in Prague, Czech Republic on Oct 30 - Nov 2, 2018.
Devcon is organized and presented by the Ethereum Foundation, with the support of our sponsors. To find out more, please visit https://ethereum.foundation/

## Transcript

those of you who are here last year and happened to come to our talk might recall ether meant a 1.0 well we're back in ether MIT is back but this time it's better so I'm gonna talk about ether at 2.0 which combines the best features we think of aetherium the EVM and the best features of the cosmos SDK and tender mint in this presentation first I'll give a quick overview of cosmos explain the network model the topology why we think it makes sense I'll talk about the feature set of the cosmos SDK in particular and how it can be combined with the EVM and I'll give a quick background on tournament consensus and how it facilitates this model of interconnected blockchains then we'll talk all about Ethernet 2.0 I'll outline the architecture the chains will be made from ether mint 1.0 and why I'll talk about how the EVM and the SDK can work together to give you native code and EVM contracts all in one blockchain I'll talk about finally how you can move your DAPs or move parts of your jobs on to the ether mint chain while preserving seamless UX and I'll take any questions so first the cosmos model envisions a network of interconnected block chains in many ways these block chains will be application-specific they will do lots of different things but talk to each other in order to facilitate this we need to make it really easy to build an application-specific block chain to do that we've built the cosmos SDK the cosmos SDK is a framework for building custom state machines it contains all the glue library code of blockchains AVL trees message handlers all the logic to knit modules together super developer friendly it's written in golang which many of you are probably familiar with it's all open source on github the Kosmos sdk is designed in particular to be secure to make it easy to use different modules which do different things which can interact with each other but where you can control the ways in which they interact so you can mitigate unintended consequences to do this we use something called least Authority permissioning so you pass around keys from one module to the other keys the control which stores can be read or written to unlike for example a call between contracts in the EVM a call between modules and the cost that's decay can only execute reads or writes to the stores for which you give it permission to do we're trying to model the SDK as a sort of Ruby on Rails for blockchains a way to take existing modules add your own custom flavor for your particular application and end up really fast for something that's production-ready in order to do this the SDK needs to be modular we've built a system by which we can build particular components that you might want in your state machine as modules now and eventually not only us we hope but many other people who are building these application specific block chains they can build modules as we've built for governance staking fee distribution and inter blockchain communication they could build a different module for add X or a different module for a different kind of proof of stake algorithm then you can using the cosmos SDK pick and choose which modules you want to use in your blockchain glue them together and connect them using the least authority permissioning system to run your state machine we hope then when you've done this and when you've added your application specific flavor the custom module for whatever your blockchain is going to do better than anyone else you will share that or share parts of it upstream so that people upstream and downstream so that people can build upon what you've done improve it and send it back in order to facilitate this network of interconnected block chains we need a consensus algorithm with particular properties so we've built tenement tender mint is the state of the art generic Byzantine fault tolerant blockchain engine it encapsulates the consensus in peer-to-peer layer coming to consensus on which transactions ought to be applied in which block and when a block needs to be appended to the chain and it provides vertical layer one scaling for your state machine superfast two-and-a-half second block times ten Dermot is fully peer reviewed and it offers one block finality as soon as the block is confirmed it will never be reverted most importantly for this model of application specific block chains tender mint is generic it can connect to any state machine you could write using the cosmos SDK or using a different SDK any kind of state machine you like operating whatever sort of application you want to operate and run it with ten durmand so here's a picture of what the tenement stack looks like so you can see this in visual form the large green rectangle is tournament consensus and each of the circular disks is a validator participating in the consensus algorithm sending messages to each other to vote in BFT rounds then in smaller blue rectangle we have the ABC i-- allocation that's your state machine written using the cosmos SDK and perhaps featuring the EVM which is talking to and from ten Derman telling ten Dermot which transactions are valid and executing the transactions once they've been confirmed in a block so in order to make it as easy as possible for developers to start working on full function block chains using the SDK intendment and in order to build the particular block chain we're working on the cosmos hub we've built some core modules which I'll talk about now the first module is bonded proof of stake bonded proof of stake is the proof of stake mechanism which will be used by the cosmos hub in which we invite any blockchains in the cosmos ecosystem to use themselves with their own staking token if they like bonded proof of stake has delegation as we say with skin in the game so delegate errs are liable have their valid at our commits a protocol fault they validate or double signs or fails to participate in consensus for too long the delegate errs too can be slashed bonded proof of stake addresses the nothing at stake problem with a nun bonding period both validators and delegate errs when they want to participate to vote in consensus have to lockup tokens for it will be three weeks at launch for the cosmos but for a configurable amount of time and as long as those tokens are locked up if an infraction is discovered they can be slashed we've tried to build this module module to be batteries included but extensible you could import it as is and have a fully functional proof of stake algorithm for whatever state machine you're running or you can modify it you could add custom slashing conditions for different kinds of application level guarantees you want to provide or you could change the way we've parameterize our slashing and staking conditions all total object of delegation or punish people differently for committing protocol faults the second core module is governance like our proof of stake module government's design is designed to be pretty flexible so it will work for lots of different blockchains our governance module has different kinds of what we call governance proposals proposals to change some way the chain works the first are simply text proposals they don't do anything in protocol in band but rather allow for the network to come to social consensus on maybe some change to make in a future software upgrade or on some way in which we want to change the way we're operating this blockchain the second are promising parameter change proposals which are inbound so if a parameter change proposal is passed that might change the amount of steak which is slashed for double signing or it could change the annual network inflation rate third and finally we have software upgrade proposals which actually upgrade the state machine which is the chain is running so upgrade in the case of your application specific blockchain may be the logic of an exchange once everyone has come to consensus on that proposal at a particular block height automatically for governance we've elected to use liquid democracy so in tender mint block chains there will be a few hundred or so validators those validators going to vote and by default they were carry their delegate errs vote with them if your validator votes for in favor of a proposal and you as a delegator elect to do nothing your vote your stake will count in favor of the proposal but delegate errs can also elect override to vote other validators if they disagree and they will have time to do that if they see what the reality to votes for and decide they don't like it next up on the list of core modules is fee distribution rewards and fees we have decided that we need to use a pretty different system for fee distribution for tender mint than for most existing block chains we can't just give all the fees to the proposer because it would make it too much of a incentive to bias proposer election or das the current elected proposer so we just distribute most rewards for posting proportionally to stake held by all the delegate errs and Vala leaders on the blockchain like the staking module and the governance module our fee distribution module supports any token so you could integrate it into your application specific block with your own token easy peasy just changed one string and it also supports multi coin fees so we think the economic module model approves take may be a little different for example for the cosmos Club we plan to accept transactions or we plan to encourage the validator set of the cosmos hub to accept transactions which pay fees and lots of different denominations they don't need to be atoms the final core module and my favorite is inter blockchain communication so this is the key part of what makes the cosmos network model tick we've exposed the IBC interface which is a payload agnostic way of sending authenticated messages exactly once to another blockchain it's a little like tcp/ip for block chains on top of this authenticated exactly ones message delivery primitive you can implement lots of different application level guarantees you can implement token transfer which means that you're conserving supply across two different block chains you could implement transfer of unique tokens which mean you're conserving the existence of the particularly unique token on only one chain in only one account at a time you could transfer data or you could even transfer code parts of your smart contracts in some kind of sharding then the special module for this presentation is the EVF so in ether mint 2.0 we've taken the e via in the etherium virtual machine which is the core part of the etherium block chain that runs all your smart contracts and put it in a cosmos SDK module has all the features you're used to an account database and a state tree it can run aetherium transactions that's how we've been testing our EVM implementation by running the actual theorem transaction history but most interestingly it can call to and from cosmos SDK modules so if you have a block chain with the EVM module with the governance module with the staking module contracts you write for the EVM module can talk to the staking module maybe they could delegate maybe they could query balances maybe they could see when people got slashed or do things when that happens similar for governance for fee distribution and for IBC so if you have an EVM contract on a blockchain in the cosmos SDK along the IBC module contracts running on that EVM module can make calls through IBC to and from other blockchains in order to enable this we need and we have implemented a shared state view so there is one token or only needs to be one token on this blockchain which will be either quote unquote in the EVM message dot value for those of you familiar with solidity but which will be the token used for governance possibly for proof of stake and fee distribution and the other causing assist eke eight modules so ether meant the whole package combines the cosmos SDK all of these core modules we've built the EVM module and a web 3 API layer we think this provides a pretty compelling deal you get the scalability of ten durman's Byzantine fault tolerant consensus 2 second block latency instant finality the power of the cosmos SDK lease capabilities security the ability to integrate all these modules and contribute downstream and the existing ecosystem of etherion contracts and development tooling truffle metamath solidity you name it but we've even done more than that not we have not just poured it as an ether mat 1.0 the evo module from death we've ported the Evo module from turbo goth which Alexei is working on that is features a lot of improvements over the base EVM it's faster we've changed up how the database works of course it connects through the SDK module interface so you can have what are like pre compiles in the EVM but which really call in to other cosmos SDK modules or vice versa Turbo goth I will encourage you to take a stop at Alexie's talk which is tomorrow afternoon same time but in brief a turbo gap which is being developed in parallel as an aetherium client along with the cosmos sdk module has numerous improvements over the current available EVM implementations it supports any binary search tree has structure preserving custom serialization as opposed to our LP which is much faster it allows you to use the same data structure for database indexing and merggle hosting also faster and it implements a new caching strategies in particular for state sync using the benefit of hindsight so using the fact that you know what future transactions are or future states when you're trying to sink past transactions and verify them now there are two ways to use the EVM module as I've alluded to ether mint will exist as a particular block chain as a proof of stake chain with all the core modules with the EBA motto and with a web 3 interface for running smart contracts but the EVM ECHA will also exist as a library the EVM module is a library will allow you to deploy your own cosmos chain with EVM support add in other SDK modules in whichever combinations and parameter choices you like and choose whatever token in economic model you want the e thurman chain will also be launched with a term you might have heard the hard spoon what's a hard spoon a hard spoon is to copy the token distribution including multi signatures from another chain onto a new chain so in particular for a thurman were planning to copy balance subject to approval by cosmo sub governance to copy balances from the etherium chain possibly also atoms from the cosmos chain and use that to seed the photon token which will be the native token on a thurmond ether meant we'll have we plan shared security with the cosmos hub meaning that anyone who wants to attack either mint would have to pay the combined cost of attacking two block chains of course this will all need to be approved by cosmos self governance so we're focusing on building the software to enable heart Spoons in any configuration you want or maybe even a hard sport which dices upstate and rearranges it but we do expect to propose this particular plan and we expect photons will be accepted as a fee token on the hub which is super important for user experience as I'll get you later so either where does a blockchain is designed to be like aetherium one chain for many EVM applications at once well hard Spoon account balances so users can start out with tokens to start using the chain ether MIT will be a sovereign chain having photons you don't need anything other than photons in order to use ether meant it will also have the governance staking and sashing modules from the cosmos SDK using tenement consensus it will be fully web 3 compatible so you can use truffle metamath solidity all your existing tooling you know change a drop-down slider and madam asked can connect to ether mint instead and it will have IBC connections to the cosmos hub into future zones which will inhabit the cosmos network the VM module for sdk zones provides different trade-offs so in this case you could elect to deploy a cosmos chain which includes some of our base modules and some modules of your own and also an EVO module you could even add our web 3 interface if you want this means you can utilize existing solidity contracts as part of the logic on your new application-specific blockchain and it means that if you're trying to port an application which is previously targeted towards the EVM to an application specific blockchain you can do it gradually you can start out with your solidity code perhaps deployed to an EVM module and you can replace the performance critical parts of that code one at a time out with native code maybe all the way maybe not but we come to a problem what if we do that you launch a new application specific blockchain but now your users are split or your potential users are split some people are on a theorem they have state on ethereum tokens collect crypto collectibles and some people want to start using the new ether and chain we need a way to connect them we need a way to transfer state and assets in a two-way bridge where it's possible both to deposit if you will from aetherium on to ether mint or on to a sovereign zone with the via module and to withdraw from the sovereign zone back to the etherium main chain then you can select which parts of your logic are most improved by the new features of ten Dermott's in the sdk and put those on ether mentor put those on the sovereign chain and allow you users to retain assets or wallet balances at least for a little while on theorem how do we do this we've developed at cosmos a technology called inter blockchain communication which allows for this kind of generic two-way bridge in order to bridge ether mint to aetherium we need to separate IBC connections we need one IBC connection a Thurmond and the Kosmos hub which is a blockchain that we're launching soon and we need another IVC connection between the cosmos hub and etherium then tokens can flow through those two connections so someone can deposit from aetherium some ether that ether will head over the IVC connection to the cosmos hub and then it will be bounced on from the cosmos hub to ether mint where it can be used in smart contracts operating on the EVM module Ani Thurmond what is this IPC bridge allow for the answer is pretty much anything IBC is payload agnostic so it will take some byte string from these two chip from one chain to another or from two across two bridges from etherion to Thurmond then IBC modules on etherium and on a Thurmond will expose this primitive to contracts so that you can write contracts which transfer tokens which transfer assets which do much more complicated thing is like execute cross chain dependent smart contract logic or move code between chains hopefully this will allow adapts to put parts of their logic that are most fee sensitive or most time-sensitive and execution speed on a Thurmond and leave some parts on aetherium for existing DOPS we think like I said this is super compelling you can move parts of your contract logic overtime when you need to execute an expensive transaction you shift over to either mint and the ux4 users can be pretty seamless abstracted over and all the web 3 tooling a particular example would be porting a Dex porting a decentralized exchange protocol like Xerox or wyvern to an ether mint zone so you could copy the protocol contracts on to e thurmond deploy them to the Evo module get them running then allow users to deposit tokens over the IBC bridge or even non fungible x' from aetherium to ether mint once they're deposited to ether mint they can be used and traded in the stacks and kept there indefinitely if you like and if users want or want to have the option to withdraw back to the ethereum main chain they can do so with no permission on your part you could also implement it acts as a sovereign zone using the EVM module with the same initial starting procedure of porting the contracts from solidity and deploying them on to the Evo module on the sovereigns own but then you could add some improvements you first of all you would have no competition for block space so if you run a sovereign chain for just the Dex only your Dex is transactions would be settled there you could prevent anyone else you know any crypto kitties hit of the month from crowding out your block space and making it expensive for your users to trade you could also start pointing parts of the Dex protocol I'll probably trade settlement execution the super gas expensive parts to native code as SDK modules which would make it much cheaper and you could change more complex things you could alter sneaky mechanics maybe add new slashing conditions to prevent front-running or implement alter the state machine implement threshold decryption so that transactions are encrypted when they're submitted to the blockchain and decrypt it later so no one can front run on the Dex all of this is possible if you have a sovereign zone so what's the state of all of this ether mint is not vaporware you can find it on github development is progressing exceptionally quickly worked on by bears who's sitting in the audience and by Aleksey who will be giving a talk tomorrow we're anticipating a Developer Preview release which will have a functional or 99.9% functional web 3 API by the end of q4 this year and which will allow you to deploy solidity contracts to the EBM module and play around with either mint and we encourage you to check it out send us feedback tell us what you would like tell us what the challenge is facing this kind of application deployment process are there of course many questions we haven't answered yet we're not quite sure how contract balances ought to work in proof of stake there are lots of contracts say on the etherium main chain right now which hold a lot of ether should they be able to delegate what happens if they get slashed what happens if you deposit your tokens to to a Dax contract that Dex contract delegates then they get slashed are you responsible maybe we want to figure that out we haven't decided what the exact token models for fee-only tokens since we have delegated security ought to be will they be the same or similar as proof-of-work rewards with a declining exponential maybe they will be flat 500 tokens per hour for the until the end of all time lots of opportunities to experiment of course I just went this morning to the etherium 2.0 talk which was super interesting and I'd haven't had a chance to update this slide but it sounds like they're now switching to a version of proof of stake which will provide very fast finality which were thrilled about because it will allow even faster IBC connections from aetherium 2.0 to ether mint to cosmos to any other chains which implement this kind of entertaining protocol so first I will my pitch or ask is that you head on over to cosmos network slash DEFCON 4 fill out the survey maybe you mentioned that you came to this talk ask us any questions and that you head on over to our booth for a chance to win a sweet Cosmo City but even more importantly ask us questions about ether mint how you can use it what we're up to how we can help any questions thank you very much we have five minutes for questions when is the tournament going life when is the interment going live excellent question so we anticipate Developer Preview release at the end of q4 and then the actual deployment of interment will be up to Cosmo sub governance so that might be you if you're voting was the status or was the state of the development of the bridge between the Kosmos hub and be through main chain fully specified partially implemented you can find the spec on the cosmos SDK github repository and you can also find a prototype implementation are you planning to support inter blockchain communication with with chains that are not part of cosmos yes absolutely the spec for our IBC is already open open source and we encourage anyone to contribute and we absolutely encourage people who are not cosmos teens not using tender but not using SDK to implement that protocol how portable is your EVM can I use it from for example a hyper ledger project or something else it will depend a little on the project if the project doesn't go it will be super easy if the project is not a go will be slightly harder but still possible if you deploy a EVM module it's the idea that you would restrict what contracts can be deployed into that blockchain essentially so you're not competing for you know resources with other contracts that your users to play yes yes so you could deploy an EVM module and not allow contract deployment at all after you've like launched the blockchain which might be what you want to do in a sovereign zone or you could allow limited contract appointment to require that it be approved by the validator side or something like that sorry if I miss this but like can you memorialize the internment State on the route chain so for example I you could use like the so IBM to invalidate state transitions on the route chain is that possible that is a extremely interesting question in principle yes the cosmos IBC model doesn't have that doesn't need state fraud proofs to work necessarily but if you wanted to implement something that did or you just wanted them anyway for extra security think that wouldn't be too hard the EVM is inspect compliant so if you have a spec compliant interpreter on in solidity on the etherium a chain to work hi if I want to make the transaction between two block chains both with cosmos SDK do I need atoms oh right thank you I alluded to it and then forgot to answer that question so thank you for asking it that's one of the reasons we think the photon hard spoon is super compelling because it will make that unnecessary because if photons are accepted as a fee token on the hub and on a Thurmond then users can transfer assets over both of those IBC bridges without needing to acquire any new tokens last question I'm curious what the user experience is like when you're moving across these chains we hope that it will be seamless it's not going to be seamless at first but it's possible in principle to abstract over most of the transactions and because IBC already provides this cross chain state verification you can basically pay fees once and cover all of the transactions you