Dencun Interop Testing #29

Transcript

foreign [Music] thank you

[Applause] testing call um I guess the two main things to cover were any concerns or updates on Death Note 8 that or things people feel we need to discuss

and then Mario had some updates on Builder testing and uh if there's any updates there as well it might be worth touching on just the transaction pool stuff on the El side yeah but I guess um maybe to start with the devnet is there anything people feel we should be discussing or uh that they want to bring up so we currently have uh most of the clients working which is very good but we also have a few of the clients that are not working

uh our primary suspect is the prism and members on the scl side and everyone embezzle on the es side those are the clients that are not currently working so the problem with bezu is the proposals are not coming through just yet uh I haven't tested their latest latest but uh based on their latest main uh I was in table two make uh blocks after the nap and regarding prism and the numbers bugs they are decline things are working on a fix

and I'm currently taking a look at there we go I'm saying what's wrong right there got it and um so for the clients you mentioned they're not working is it that they're they just haven't they're not ready yet or that um they uh they tried to join us that's nothing had issues and all the clients have now machine on the devnet so everyone should be able to log in and check their own node nice see I knew the client teams uh yeah can

comment on it more in detail sure um can you guys hear me test test yeah okay sorry yeah just making sure it's working yeah so from the prism side we implemented everything and of course like we found some bugs and that's why um yeah that's why it's now working so the issue that we're seeing is that when we're syncing that somehow the node gets um stock or or basically stopped and yeah I was trying to figure out what's going on so we suspect a deadlock with the blood processing but yeah we should have more update in the next hour got it

anyone else want to share updates from the time teams sure um basically real quick we're not quite sure yet what's going on with proposals and we're still looking into it so between that and um we have our 4788 implementation just about ready to go and we kind of need to get that all merged into main so it's investigation's still pending for Aragon I haven't checked the what exactly like I know that some arrogant things some blocks are bad I'm not sure exactly why I haven't looked into it yet but well like that's just my like thinking of the top of my head it might

be related to if for 788 2002 uh which kind of it's not nudged yet because we are still waiting for it to be finalized but later this week I'll like I'll take a look whether it's due to that or something else and hopefully fix the problem we had an issue with the blobs pruning but we already merged the fixed and uh yeah it's it's supposed to be working now okay any other clients want to share

updates thanks to Mario we also uncovered a logical bug in which uh when requesting uh the already built payload we were not passing back the blocks bundle on that so thanks Mario for that thanks I shared the uh VR on the chat uh

for everyone to think uh White um as you just uh new tests basically just tested the versions um so that's what the uncover did developed so everywhere you if if you guys want to run I haven't updated Titan page with this yet but um I'll do it as soon as the grinds tell me that uh it's it's safe to grow because I I would like for clients to run this test locally and fix the issues before update me anything just just to not break it uh web page for everyone PC

and uh we're starting testing we want to start testing a new text pool during this devnet live stream this location thanks straight anyone else okay um if not then I think the next thing um Mario you had some updates on the

Builder testing that you wanted to share yes just three people so we are using this mock Builder on the devnet and on group doses on Fresh Empire so I'm currently updating this for the nest to build a the payloads um basically this is this is just like a server a mock server that um it's it has a rest API endpoint that basically just tells it when to corrupt some Builder payload and send it to the to the consensus client to test like uh like the failures from the Builder side um and this was last updated for

Shanghai and we're currently just preparing the updates for uh 14f so it should build payloads with uh with blocks it still requires the execution client because it requests the payload from the execution client and then it relates it to the consensus client with with any corruption that uh that the tester wants to do and uh once that one once the update is done I think we're going to update uh pertussis images cat and also that's Nexus to introduce some Builder testing uh yeah got it any thoughts

questions on that and then next uh so Barnabas you posted the there's a couple open PR still in the devnet eight spec so is the only one I see now is the uh the receipts PR is that the is that the only one that's not closed yet oh sorry there's another one oh uh yeah we have two we have one execution apis and our engine apis and one in the uh

so okay so so it seems like yeah the execution API one we should probably just merge um if people are happy with it again maybe paying oh sorry it could be a different one it's like because in devnet 8 at least like one old PR 7431 which is this update nonso Beacon yeah but now a new one like if for seven eight eight version two uh what was it yeah so that one yeah yeah so that one I want to get to after but just the first one so the execution API is PR this seems non-contentious we should merge

this one right no but but it's it's irrelevant with with uh what's 788 V2 it's like this one is becomes irrelevant yeah yeah but sorry so there's two PRS so there's a PR on 4788 which I agree becomes irrelevant um and we don't have in scoped the new 4788 as part of that eight so maybe for that one we should just remove it from the list do people agree with this basically so yeah let's start with that one so this is the 4788pr by Martin which uh which sets uh which deals with like the the nonst uh with the nods this one is obsolete given the changes

we've made to 4788 to just use a regular contract rather than a smart ad in a pre-compile um so does it make sense to just remove that one from the devnet spec altogether yes it does yeah but I think also the 2019 it includes like like um bits and pieces of uh 4788 V2 it which is not or analyzed yet right yeah and I think we should so maybe what we should do is like plan to have four seven eight eight V2 as part of devnet9 right like a full implementation does that make sense yeah yeah probably

um because also about that that one um I wanted to check whether like um whether we want to make the the address of the that um smart contract or the beaconroot smart contract configurable or not um I mean we could uh like currently in Aragon for instance it's not configurable uh and so we are waiting for the um the final address to be to be ready but we could make it configurable we just have to agree whether we need to do that or not yeah um yeah and I feel like that's probably just a bigger discussion around

478 that we need to have so I yeah which is basically what we were talking about in all core devs and like do we want I guess or sorry are you saying like the making the address configurable so okay yeah sorry let me just take a step back so like as I understood it it's like we want to potentially have the same address on every chain and so I don't understand yeah the the the link between that and making it well because I think uh in um in in current devnet eight uh the the assumption is that uh we use the that

like the the it's we kind of you have this uh 4788 V2 but with the old that with the address um the the zero x b so like the old pre-compile right and and maybe probably for devnet nine we should have the kind of the final address if okay no not like not uh um zero xB okay does that make sense to everyone I think the address depends on the

finalized goals uh so if we have finalized code for them tonight I guess I think yeah and and My Hope or my assumption is that we would yeah yeah okay that sounds good and so okay and so then for the other PR that Barnabas had so this is execution apis PR 398 add blob gas use and blob gaps price to receipts

so should is there a reason to not just go ahead and and merge this I'll post it again in the chat yeah yeah okay uh so no oppositions to merging yet um so yeah let's try and get this merged uh do today tomorrow so that um yeah we can have like a finalized uh finalized spec and I guess the last thing uh I wanted

to bring you up is is is is that the transaction tool on the El um so I know we discussed this a while back uh and and Roberto had written a doc uh trying to document different approaches um but I'm just curious to to hear from the different teams like um you know do you feel like you have a good approach to deal with the transaction pool um is there anything there we should be looking into um yeah I guess this

or maybe maybe okay this is a better way to find this like assuming that no one no one seems to have concerns about it um what's the what's the best way we should go about testing the transaction pools so I assume we have plans to do things like uh reorgs um and and uh and with not the runners and and obviously testing things through like external Builders um but other other things that we should be doing uh to make sure people are confident in their transaction poll implementations I guess I guess I can start by saying that probably we should reinforce the

transaction pool tests on hive and it's very uh small set of tests at the moment we can start by including more uh transaction problem reorce base test on hive it's not it's not that difficult it should be should be doable quite quite quite easily six and your results on that all right anything else anyone wanted to discuss okay and in that case we can wrap up and

um I guess Maybe I would keep this call in the next two weeks but uh we can see if they start to feel like Superfluous maybe uh sun setting them and just having ACD and AC DC will be enough um but let's keep at least another one um to see if we have more stuff to talk about in two weeks and otherwise we might be able to just move to the regular quartets calls and that's it thanks everyone talk to you all soon thanks Mike thanks Mary thank you