PeerDAS Breakout #14

Transcript

e hello can you hear

me hey yeah how you doing good and you yeah very well thank you hello hello hello e e

welcome to the P breakout room number 14 okay and good to see many fin faces now IDs handles uh yeah I just share the notes and agenda in the chat so as usual we started with C update I think it's okay to say um not much progress during the holidays okay say anyone want to start I I can start yeah um so yeah uh prism

now is a to to start a d and then to go to start a d and to go to Electra and then to Fu and with activating the activating pet the food for EP um PR is also able to think uh how to start a n at Fu and to think from scratch or from the name via Electra to flu uh one being piece missing is I still use custody sub count and not custody group count I'm working on it today I think it should not take too

long and what once it done normally I think I will be able to start interring with other clients if really that's all for me thank you Manu and may low star uh we're also working on the switch from um uh to groups and uh not a lot of progress over the holidays and we're also switching uh uh implementer so just catching up and

uh more update next call thank you Matthews and Nas yeah so we are done with custody groups um fully tested it with the test vectors that were generated recently I think they're not yet included in Alpha 11 because it hasn't released yet here and um I think we have some bits uh left basically column syncing in fulu U some part of that is left apart from that we are um we are like perfectly fine with pias at fulu so like I don't think there is anything else left to be implemented

from our side except uh column syncing from F because custody groups is done yeah so that's that's pretty much it from ners thank you and uh Lighthouse hello everyone um Jimmy Li house um yeah not much progress over the holiday period but um we're back um and um we've got a um the coupling subnet working so we got custody groups um that's working and we've got local de net working as well and we've also implemented the g g blob

side car h P API and also fix a couple of small bugs um and at the moment um we also working on checkpoint syncing um as well as full Activation so we don't have full activation yet um we're still working on the boiler plate but it should be ready in a week or two and we should be able to interrupt soon um that's all my house thanks thank you and te and not much progress on holidays uh but uh we have a full branch uh we need just to polish it a bit and after that we

will uh work on decoupling subnets and introducing custody groups that's all thank you anyone from gring is here okay um then we move to the def update and so I think people are still working on it but uh we I know we have one at least uh one big discussion that could be the Blocker of defate you could start from

it um so then you edit it to the genda would you like to give a brief of what this PR about yeah sure uh just in the current full spec uh we introduce a new blob side car by rout and uh blob cyle by range V3 um but I think it's not useful because starting at fulu we won't um I mean we won't communicate with the peer-to-peer request between ndes using blob S car only thata column s cars I agree that we should not remove the blob side car RPC

defined at Electra at least uh at least at least during the window rtion period because if you start a new no you want to sync uh some blobs before full um and so yeah the spe just proposed to remove a blide car by route and by range vory so simply that any suggestions of me or like objections or supports to

this p on C things so I I had questions about I guess the how to say intent of of of having both in general because there are a couple of ways of of implementing this broadly one is to store both the the blobs the decoded blobs and an sort of on dis in parallel with the columns and then the others just store the and then sort of decode um and then just fetch them off disk uh whatever database format something

another is to Only Store The Columns and then decode on uh recr request as that and and and my concern I guess about the ladder generally speaking is that it puts a lot of work that's not entirely trivial uh to do this decoding um on on the client responding to the reest request which doesn't exist right now that's kind of qualitatively new as it's a database hookup currently it's it's not computational so I but I guess I what I Wonder has this been what what is the general intent here I guess is the question do you mean what is the general intent to remove the vory or to

introduce vory um well to have to be able to to act to what what least had be the intent to offer access to fundamentally the same information at suevel but through two mechanisms one a colum mechanism the other a blob mechanism like but but simultaneous yeah but one thing is um so only basically super or node cing at least half of the columns will be able to uh sell blocks blobs sorry and standard node I mean full node without any EV validator will only

have four columns so it won't be able to serve any blob and it won't be it won't be able to to store blobs I agree for supernes it's a possibility to both store colons and blobs or only store colons and reconstr blobs um when it is needed but uh I don't see really the advantage of doing that I mean um if you use only thata colum subnets it will work pretty well and I don't see any advantage of sometimes for super ND uh using blob col blob blob

Subs no I agree to be clear oh sorry I I want I I do agree with I sorry I will say I support this PR so okay start with that um but yeah okay that I guess that was the that was the question that makes sense anyway I think um I think the plan was to activate pias at a pias for T later in sometime in fulu so we would be spending some time in fulu where there would be no columns but we would have a raised Max blog count and we would have like we would communicate using blob site CSE that was my like thought that I was having but then I saw there was this

thing which said that blob subnet was deprecated so I didn't know why the V3 was there in in the specs even like in the first place yeah all right okay um so we will include it to the next incoming defate

right yes thanks uh I'm I agree with that thank you um any other like defy blockers and any issues or def we want to discuss today Jimmie um do we have an updated spec of the dep net I think I think we have a page on Deet spec before um is that still up to date maintained it in the last week so I can try updating it is it um Alpha 10 plus the new pier moving V

tree I think there's no much no not many vs uh was merged after after 10 but I'm checking out yeah I think there's at least probably fullu activation sub coupling and then and then Manus new theing is in al 10 spec then that's a bit interesting now is um so assum we

cut this this week for this i p death items but um the tchra definit would be uh would stick to the the alpha 10 version is that okay um to carry yeah so I would stick to the existing release for the Petra DEET cycle and we use the new one for the P Cycle yeah that works I'm sorry is there any Electro change opetra changes in the new in the upcoming release fi 11 or just p uh I think that would be for sh

right I don't think there's been any I don't think there's been any breaking changes in the spec so even if we were to update pectra to Alpha 11 I don't think it would make a difference uh I can just quickly check that though oh the think only some non substantive item what about EAP 7742 that's been the mve TR yeah but is it planed to be reintroducing Alpha 11

why would we reintroduce it I don't know it's it's just a question no no it's been completely removed um off the table for Petra pias for everything okay okay perfect yeah at least just quickly looking at the um consensus specs commits I don't think there's anything that would break the just some test updates and minor changes so I think we can update all Petra as well as pias to use um release Alpha 11 can we agree on uh three changes for

alpha1 in terms of pias one is um activating pias at the fu Fork Epoch um the other one is getting rid of Max blobs per block Hulu and the third one is the V3 of R crisp like can we all agree to that like having these three changes in Alpha 11 or the deet yes fulu Fork was already in Alpha 10 but like um the way the blob situation and the V3 was there it seemed like pias would be activated some time

later not at the fu for like just need a confirmation there at least the way I implemented it in prism is activating puras at the full for and is what we decided I guess in the last uh breakout session yeah that's the that's the best yeah that's that's the same for us yes so there is no more not anymore EIP 7594 for K yes yes yes activation s activation nwor oh it looks like the valid

validated custody Pi is also in the de net for specs I I think want to remove right yeah I'm just updating the specs right now and I can remove well it's marked as closed um I will just make it explicit it's not there thanks about valid validator CL I guess um like I said the last time if people want to implement it normally it should be okay I mean it's it's okay if some clients implement it and some other client does not implement it because it does not break anything

[Music] um I guess maybe we can say yes if client want to implement feel free to to implement yeah but I think the pr itself has been closed right so if a client does implement it it it's kind of oh no it's open okay um yes I will leave a note that it's up for clients to decide so if we having the next beas devet uh in like in the next two weeks or so like more than greater than equal to two two weeks then I guess uh we can nimas can have validator custody um has anyone checked if there

would be any weird edge cases if some clients have validat custody and some don't yeah um I mean normally it should not because um it's it's like it's like if the if cents and Val are custody it's like if more client will be like super n implicitly um but the issue uh maybe if people Implement validator custody in a death net maybe it will implicitly make all the ND super no um which can be an issue I don't remember how many validator per Bon node

we have and I guess the limit to be a super is 64 I'm not sure about the figure but if it is the case that may make all the bon no super node and which is not something we want to test so that's configurable so I can still um edit the kosis config to make sure that explicitly some nodes have just one validator or something okay sure okay I think I updated the pet for specs page can someone have a look and let me know if PRS are missing and if they are just post them on chat and I'll add them

to the spec sheet and does anyone have uh an objection to just making this light all so I I think the pias group is still relatively small so it would make sense if more of us are actively updating this and just for contacts usually babus is the one updating it but he's on vacation for at least another month maybe two months so more help would be appreciated um I've made it right

everyone's so you should be able to refresh the page and edit and any other p uh te needes

discussion uh yeah the only thing I had was um if clients have a branch that they're able to verify that they can locally run um then just post it on the podas testing chat and I can already start running some interrup t or updating the CI and so on but yeah I guess you guys want to wait for the consensus back release for that at least so that everyone's on the same page thank you Perry it seems I canot sign in with my GitHub

on HD uh I don't think you need to sign in at all you should be able to edit the page without signing in any other questions okay uh then we move to most more space discussions um is there

any or issues of the space we want to discuss today nothing then is there any open discussions we want to discuss open topics K hello hi everyone yeah I'm back to

Matrix and I would like to ask Jimmy what are the issues with curosis uh because on the teos site we also like have some issues uh do they correlate I wonder what what happened with curosis you you mentioned that updates broke okay yeah um it's I think it's just the the ethereum package now supports it no longer supports EIP 7594 for Epoch um the configuration item um so that's why probably why teu and Lighthouse is not activating pias because we still use the Erp 7594 Fork EPO configuration um and that should be fixed once we Implement uh Fu

Activation so uh from from de and Lighthouse metrix will back only on full right mean p metrix yes um once we've implemented that um if you need to test metrics um you I guess you could try um etherum package 4.4.0 um that's the version I've been using but hopefully we'll get to very soon okay got it um I think Nishan is not here today so we still didn't come to agreement about uh milliseconds or seconds probably I will um pack him in the pr so this is the pr we have a new updates and

discussions here please other clients welcome to discuss as well thank you hey thank you Kya any other feedback back suggestions okay um there anything we want to discuss before we close Okay um thank you everyone for attending it we will have a happy peer

death 2025 thank you thank you bye see you guys goodbye