Ethereum in next two years: Prague & Osaka hardforks - Dragan Rakita | Paradigm

hello everyone oh nice uh uh yeah my name is Dragon I'm working rat I outed rvm and yeah I want to talk because the current scoping of the prag the next half is happening right now and I think it's very interesting topic to talk about after that there was even talk what to do after the pr that was like on the level of the next two years so I will cover that and some bigger changes that are potentially coming after Prague uh before that um etherum governance ah this is better the number of teams over the years like increased with consensus L clients coming in the scope of the ethereum so the governance that was like working few years ago it's very hard to make it work now the problem is there is lot of like uh research topics lot of uh like different ways that etherum is evolving so for the next like few months maybe year there will be some evolution of the Ean governments they will still be like Loosely based so it needs to be uh basically idea is to engage broad Community but proposals are mostly to uh allow cev calls to be more high level discussions where the for specific topics we will organize breaking groups there where specific like questions and more like broader um discussion can be made other than that asking communication is something that was mentioned a lot by uh multiple teams where decision making and everything was a lot easier when it's writing and when you can aggregate the the stances of multiple teams and there was a need to have more structure a inclusion a is a way to include change to ethereum currently there is just one stage consider for inclusion and after that it got included so the the time from the like making CFI to inclusion is needs to be more specified uh this is link there with more uh discussions and more like summary of how that could evolve Prague and Petra Fox prag is for the execution layer where uh Petra be is for the consensus layer there's two far Forks that are basically synchronized to happen at same time because they require both changes from both sides expected delivery for that was uh end of this year of and depending on scope could be Spilled Out for the q1 of next year this is list of aips that got included uh there is seven eight of them and uh additional two are added or basically soft uh talk uh they are added as a soft Fork after the prag and I mostly talk about them uh in ethereum interrupt that was happened that happened like two weeks ago uh we um we created devet zero with all those a implemented and it was very successful launch of some devet and seeing that all the clients implemented some version that can be run on on basically devet uh first AIP is PR of pile for bls2 uh BLS 12 is something that's used to aggregate signature on scus layer it was uh removed for one of the past Fork because there was some problem with consistency between implementations but it's finally happening inside uh execution client uh there is nine preop that aggregate all functionality that you need from BLS AIP 2935 is saving history Block ases in State uh currently the evm has block hash op code that allows you to fetch the block hash of the last 256 blocks the problem with that uh block hashes are found in the headers so if you want to make proof of the past parent block you need to for example block minus 100 you need to have the headers of all the parent block in the chain to make that proof valid including Block hashes in the state that proof is just like proof of the storage this is done as the system call to one preloaded bite code that's going to be uh created uh basically in as ordinary Creator section from ordinator Creator section and the system call is called before the execution of the block so the parent block is included before an intersection start to to be uh executed there was few changes small changes block hash was first itach was for blockage to be changed to have more gas to be executed but have like access to more block hases but uh we didn't like want it in this hard for to make it more complicated so block has stays the same and all all the block has can be accessed with the ordinary call up code uh these three one are related to connection between execution layer consensus layer first one is the platform basically it l specifies how that um communication is going to happen it allows us to be more like hey if you want to communicate from execu layer to con layer how that would usually work the second one Supply value to deposit on chain uh allows uh its simplification of the prot protocol it allows um consensus layer to be to easily fetch uh deposit from validator uh why that was not done previously because previous we had the merge and before that we didn't have communication between those two layers and consensus layer needed to itate basically find the filter all the logs before finding the deposit contract and logs that the deposit contract basically uh transmitted uh this is simplification where execution layer is going to filter those logs and just send it to consens layer uh third change AP 70002 uh this allows execution layer to trigger withdrawal for validators validators have two keys one uh Co key and hot key call key is used for the staking basically for you to stake the the the itat and hot wallet is used to sign validations sign blocks and everything around that only way before this is for the hot key to withdraw it's basically uh trigger withdrawal for consus layer this change allows execution layer to use cold wallet to trigger the same thing uh increase max effective balance this allows validator to stake more than 32 ethereum heard from consensus layer clients that this was harder to do because you need to propagate this change it's not just ass assume 328 it's like maximum of 2,000 so you can stake from 32 to 2,000 e in one validator and the last one uh was most related to sens layer and I think it's simplification of the protocol so it's a little bit more performant but yeah I'm most focusing on execution layer so I'm not sure how uh what exactly is going on there account obstruction is something that needs to happen on eum in some way or form uh on Deet zero we included 3074 that add a new two up codes that allow uh ordinary accounts basically externally owned accounts to be to be used by smart contracts uh while the uh 374 3074 was implemented new proposal arised 77 702 that basically allows smart contract to be added to transaction and executed uh before in the place of um externally owned account there is currently discussion but how and basically what we want to include uh account obstruction is some way needs to happen on ethereum because unlocks wallets and allots new like usages on ethereum uh changes are very small to implement the bigger question is what exactly do we want to include there is breaking session like happening every week on account account obstruction Pras is something consensus layer related that is uh basically included in the fork after the prag basically last year last week it allows basically data availability on the Block blob data that means that not all blob data needs to be transmitted over P2P that means we can increase blob data count blobs count and uh don't override don't um basically make P2P Network unstable uh ethereum object format uh it has long history it's uh group of small aips that improve the legacy evm evm is here for like 10 years and there will no breaking changes that where the op course got deprecated or removed there is some uh for example self destruct was repurposed for something else but removing op codes from Legacy contract was not possible uh EF introduces B Cod format not very versioning among other things uh it allows uh removal of the gas and code observability that means the contract doesn't know about gas that allows us to change that gas in the future without worrying that we are breaking some contract same with the code observability we could have code upgradability where the because execution basic smart contract doesn't know about code doesn't know about hashes doesn't know about code size we can switch it for something else and in the end uh some measurements show are showing like it's for slate it's 5 to 10% Like less gas for some cont contracts the end it's maintainance IPA that fixes some things that will long overdue on evm you're probably going to talk about that tomorrow in allore Dev about inclusion in the Prague if the clients want to include it or not that's mostly about Prague scope of the Prague is around there there's still some uh client that want to include something want to exclude something uh it's still uh scoping is still happening but most of aips are already the like even implemented and decided to be included after PL prag uh that was uh end of 2025 maybe 2026 we having new topics that needs to be addressed uh verle is one of the those verle is right now soft approv for Saka basically for the next for the next Fork after Prague that means that clients usually all started working on it or the planning to work on it verle is a big change it requires basically switches Merkle tree the work tree it allows smaller proofs it allows to have witnesses that are small in size but even transition from One Tree to another tree is very sensitive and very aopr so it needs to be very tested uh verle in general can be split maybe in four four stages one is replacement of the Merle to the verle second is transitioning because we need to have some way freeze Merkle tree and transition while updating State rout uh one of things the verle does is include the B code inside the tree that means we need current Mer tree includes just the hash while the work tree wants to include Parts basically chunked part parts of the B code inside the tree that mean there is possibility of increasing bite Cod size while um creating witnesses that are very smaller just Witnesses on part of the bu code that got executed and work Witnesses is something that will enable light clients uh stateless clients that uh basically you could had client that will take block with Witnesses and verify that block is correct uh a for fors uh this something that needs to happen in some way or form maybe is going to happen the change to protocol to enable for force is minimal but it requires structure that's going to support dep application of the blocks uh dication for FS are dication of the history blocks receipts intersection everything that's I don't know uh 10,000 blocks older than the tip can be removed from the protocol that means changes for the P2P the problem with that you will need to have ability or have a way to store those history blocks and without that storage mechanism uh it's un it's not great to just remove it portal is one of the solution that comes to the Mind where Portal is um U basically allowing to store the blocks history and everything around that uh spread throughout throughout the clients and allow one client in the portal Network to fetch the data that need it's one of the ways where we are going to store the blocks and it's probably going to be included Incorporated the clients and used in that way uh last um the big not problem but one of the research topics is how to be sensorship resistance how to have it like inside the protocol currently we are having map boost where the blocks and reers and the proposers work together in hog protocol that may they're basically building the Block in that way it works right now but if it would be better if that protocol is incorporated inside the ethereum uh that e inside PBS is enshrine that means like enshrining Pro uh proposal Builder separation there's current breakout session on those topic how how is what's the best and what's how to include it easily inside eum inclusion list is one of that that basically force uh next block to include list of the section that needs to be included uh yeah it needs to solve uh basically timing problem because currently the block Builder wants to push their block building at the last second and propagate block after that that that no D destabilized the network because the timing games and everything around that pushes it the brick part of the network receives the block part of the network doesn't receive it and yeah I think that's it just say notice uh there is a lot of topic research data done by multiple teams so I'm just having a like overview what's going on and what's like hot topics that needs to be addressed in next next two years um thank you for listening if you have question than you [Applause] good hey um I want to ask you what are the biggest um challenges regarding PBS uh having good solution that Compass basically solves the problem because you you're dealing the basically F memes uh you're dealing with uh sorry it's on my screen you're dealing with timing games and you you need to have a weight to slash the Builder that's not um they are basically not doing great for the network currently this is not done every the everybody trusts the Rel layers everybody assumes that Builder is going to be okay and if needed it's going to be removed from your layers that kind of protocol that's currently map boost needs to be moved inside the ethereum it's lot like research needs to be done and we need to be sure that everything works to be honest I'm not investigating a lot about that topic so that's my view of it hello uh thank you for the talk um do you know if in Petra there will also be the inclusion of this uh collocated um staking penalty uh not that I'm aware of you know which one I'm talking about right the one that uh tries to um incentivize St a bit more by um moving the curve I think it's uh yeah but this is not something that's going to be included at least from my perspective it's more like consens layer thing but I think that incentivizing the moving curve that's valid it is going to be rewarded the I it's like separate theapy what else say if ah there's more questions hey um sorry I was a couple minutes late so I don't think you mentioned it though do you think the contract size limit is ever going to be removed um do you think the contract size limit is ever going to be removed probably with TF if the UF is going to be include you don't have like restrictions on the code size you have like different code sections that can be basically I think the you have 64 code sections maximum something like that no 256 and it's like can be quite bigger we're currently keeping it at the current limit because we don't want to disturb what all these there but yeah this one of the things that could potentially happen um what the le5 for um e uh basically you you remove jump destination you you moove analysis you remove other five would be hey B Legacy bitecoin is Legacy bitecoin there is lot of bad things there not bad but it's like Legacy for reason people 10 years ago didn't gav would didn't didn't think that Hal can be dis usable in different ways so fixing those small things and packing inside thef is basically from removal of the gas limit from the calls basically there is different kind of calls that work the similar way uh from like removing adding immediates that basically allows op course to have immediate value for example think about static jump with immediate that's going to be uh where to jump and there is a lot of lot of those small changes that pack together the sub routines are something that are added and uh SD truct is removed and yeah uh thank you very much if there's no questions guys just a yeah quick one before you go so for those of you that are attending Russ Summit tomorrow I thought you maybe give them a little insight a little sneak peek to what's in the store for them I'm not going there yeah you're not going oh God the rust OG is not going what's happening no I'm I have different like okay yeah person life things yeah got you man it's been a long three days but another big round of applause for Dragon guys thank you