All Core Devs - Testing #78 (ePBS Breakout)

Transcript

Alpha 5 specs are going. Yeah. Hello, Pettis. You found it. Uh, is there is there anyone from Lighthouse here? >> Yeah. Um, >> did you get >> we have a we have a alpha spec 5 branch that will probably be merging into our trunk parent uh pretty soon. Uh, we're passing I believe all the spec tests. I saw there was a broken fork choice test that you guys fixed. I thought we were passing those fork choice tests. So maybe maybe we're missing something. But yeah, um we're we're in pretty good in a pretty good spot. We still need to review the the PR, but hopefully in the next day or two it should be merged into our um trunk branch.

>> Awesome. Yes, I did merge that fix from Nico. I can share a link here in just a minute. Uh and great um prison >> Terrence is not here. No. So Terrence has a a branch uh where he's implemented everything balls and the the slot number and everything and uh uh we're reviewing it. Uh we're hoping to start merging it to trunk uh today actually as early as today.

Uh I don't think we're passing spec tests. We're skipping them and I'm unaware if uh Terrence already has checked the latest ones but uh he told us that his the branch actually passes them on on the on the tip. So we can only find out on on develop once we merge. >> Okay, cool. Thanks uh Ter. Uh yeah, so I'm working on a PR which passes the reference test apart from fork choice which uh Enrique is working on fork choice changes. Um yeah, so I hope in the next couple of

days we have some branch that is working. >> Great. Uh great. And I saw your comment in the CL testing uh chat. uh about deprecating the execution payload header SSC static tests. I will look into that. I think we can um I would like to deprecate uh sorry list out the deprecated containers and functions and everything so we don't generate tests for that and that would be part of that.

Oh >> yeah, thank you. That would be good. >> Yep. Um uh Nimbus I guess someone from NASA is here. So uh we don't have a single branch that passes the alpha 5 spec test but I did experiment with merging a couple branches from all the open PRs and they do uh pass all the spec test including the fork choice.

So we just need more time to merge the uh open PRs into unstable. >> Awesome. And uh thank you for your PR with the test fix grand. >> Yeah, we started to work on on this um five test will take two days I think. >> Okay. Um, does anyone have like a crytosis setup with an EL doing a local I'm not sure if Prism should be ready to

go. Uh, but we have the E image zero from GU. So if you're testing, you should be using that image. that that has like the slot number and val stuff as well. >> Yep. That has everything. Uh but ballnet three spec with the 770 PR. I'm going to do testing prison uh later today uh because uh Terren uh just

paying me deep branch. So I'm I'm building that to I'm not sure if that should actually pass everything. Uh we also have to update our tooling. Uh I'm just waiting for a client team that is uh reg before we can update our tooling with the alpha file. smart of us. Uh is there anything else we'd like to discuss issues? Uh anything else?

>> Yeah, it'd be nice. Uh so I was contacted by Matti uh from Panda Ops. Um, it probably not the full-blown like sort of like attack scenarios, but it'd be nice to be able to set up a Corttosis run where we can test the typical things that broke us. Sort of like checkpoint sync uh syncing uh in the presence of missing payloads uh checkpoint syncing from uh the missing zero slot and these kind of things. So that at least we know that uh if the devet goes wrong we can restart clients.

So this will require from Panda Ops to uh to have >> you already sent him all the scenarios because >> no no the scenarios that I would discuss with Matthew would be uh we haven't scheduled anything yet would be most uh would be more serious scenarios of of attacks uh or or sort of like more serious things like what what we found this bug on Prism uh in the case that there were blocks epochs of missing blocks but uh but the common The common more more common things are a payload was reorg and we're trying to reync or a checkpoint sync where the last block of the epoch was uh 30 slots 30 with and without the payload and so forth. This

these sort of things should be uh tested. If you want I can give you a list of those. >> Um I think the list would be great then we can work on it this week. >> Great. Yeah. I can work on that now. Ideally, we should have the prism truck branch ready to go with the previous changes because uh I know that that's what he is waiting for. >> Yeah, I'll work on the list right now. >> Would these be created with like a tour? Is that the plan?

>> Probably. Yeah. Okay. >> Yeah. And there's also a fork of prism we maintain that we can add some malicious um malicious stuff too. Okay. Anything else from anyone? >> Okay, bye everyone. Um, talk next week and see you all in person.

See you guys. Thank you. See you. Bye. See you.