RPC Standards #36

Transcript

Mercy Boma Naps-Nkari: Hello?

Hello, good afternoon. So, we can wait for, like, the next two minutes for Tuesday.

Hello? I'm not sure if I'm audible. No. Okay. Thank you. So Chase said he would join in a minute, so I think we can wait for him, since most of the issues we have, it's on his meta-tracking issue. I don't know if we can discuss about, error codes today, if there is any news on that, Sims.

Simsonraj Easvarasakthi: Hey, everyone. I tried to reach out to Felix, did not get a response. I was hoping I would join this call. I I Dm'd him, and we also have a thread in the Channel. Yeah, I'm planning to just go ahead and do the implementation. maybe get the PR rolling and have a conversation in the PR, maybe? That would be better? I'm not sure, because, this is taking too long. I don't know if anyone else have anything to discuss. On the error codes… Personal opinions. Okay.

Mercy Boma Naps-Nkari: I think like the best way to go about it now is like to open a PR and then. The PR should kind of, like, force a conversation if necessary.

Simsonraj Easvarasakthi: We have a couple… actually, at least two, one done by. Chase, and another one from Sina. There are a con… there are a couple… yeah, there's been a huge conversation in the same PRs, but we haven't come to a conclusion. So I thought we will have… we could hash it out here.

Mercy Boma Naps-Nkari: Okay, the main issue is, I'm not sure I fully understand what the blocker is, and then… I don't think.

Simsonraj Easvarasakthi: soon.

Mercy Boma Naps-Nkari: I can try and.

Simsonraj Easvarasakthi: Rice. So, yeah, the main issue is that, the, spec, or the new error codes propose Newer values, like mostly positive and, groups. That we've all discussed, so many times, but the simulate V1 endpoint. uses a completely different ranges. It doesn't have any grouping, it just uses, I think, from… Let me see… yeah, minus 38010 to minus 38014 in order, in some sort of order. I can post it here. This is the difference between what the spec proposed, plus catalog proposed, plus what Simulator uses. So the main confusion or main blocker here is, how can we unify these two? The concern is every other method will use the newly proposed catalog. except simulate V1.

Yeah, that's that's the concern from some folks I've heard in the thread. Why, in my opinion… I think we should just go ahead and deal with this and simulate V2.

Lightclient (Matt Garnett): I don't think there's gonna be a way to… You know, unify those error codes without a new method. So I think at the end of the day, like, we just need to educate people who are implementing interfaces to the RPC and say that if the error code is positive, then you need to You know, handle it with this new error group. Mechanism, or if it's negative, then you have to continue using the legacy errors.

Simsonraj Easvarasakthi: I agree.

Lightclient (Matt Garnett): I think that's kind of what Felix and I talked about. I mean, this has been two or three months ago now, but that was really the only pathway that we saw. Like, it doesn't make sense to go back on V1 and change what the values mean.

Simsonraj Easvarasakthi: Yep.

Lightclient (Matt Garnett): It feels weird to… You know, re-specify, add in these extra error codes. You know, across the entire interface, like, we should just…

Simsonraj Easvarasakthi: Yeah.

Lightclient (Matt Garnett): Switch, and… I think that that makes… that makes things not that difficult for library implementers. It's a simple check. Is the… Is the error code positive? Handle the new way, it's negative, handle the old way.

Simsonraj Easvarasakthi: Yep, yeah. Okay, great. This is what I was waiting for, some sort of opinion from someone. I think that makes sense, because I'm… I was also saying the same thing. I think it's, it's not that bad of a difference, too, because, not everyone uses Simulate V1. And especially the end users, I don't think, like, wallets or cross-chain clients doesn't rely on SimulateV1 too much, at least not yet. We're trying to simplify the users and end users and cross-chain teams. So grouping definitely matters for them, so I, I agree. In that case, I can proceed. I can create PRs. This is what both Sina and Chase tried to solve in different threads. We've been having this conversation. I… thanks for the comment.

lightclient? Yeah, I can… I can go ahead with this. Let's just, create a PR, and then if anyone has opinion, we can deal with that in the PR.

Lightclient (Matt Garnett): Sounds good to me.

Simsonraj Easvarasakthi: Thank you. Awesome. Finally.

Mercy Boma Naps-Nkari: Okay, thank you. I think, okay, show my screen. Oh. So most of the issues we have on this meta tracking release, which On the last two calls, we had a blocker from Geth, two Geth PRs, and then a Hive one. So I'm not exactly sure where we are on that. So.

Chase Wright: Yeah, I mean, nothing… nothing has changed. The Go Ethereum PRs are still pending, like… I have… Approval on one or two of them, but they're not merged, and then… I I brought it up in ACDT for basically the same. reason was… That. I need these PRs merged. So, I don't… I don't have an update there.

Mercy Boma Naps-Nkari: So we basically stuck on everything on the list. Then the only new thing should be tracing APIs, which. There's been like a huge and a ton of PRs.

Chase Wright: Yeah.

Mercy Boma Naps-Nkari: Trump.

Chase Wright: I mean, I think this one is too early for us to, like… make any, like… like, it's not ready to be merged, but it's very close. I mean, Banteg has been… working on it very hard. All of the client teams have been merging in their fixes. I don't… I don't know if Geth… Team has done any… Review on their side, but all of the other clients are… Merging PRs, I think Besu currently has the most open PRs, but… You know, that doesn't… that doesn't really mean anything. Yeah. I don't know. It's all…

Mercy Boma Naps-Nkari: My question is, sorry, Sorry, go on.

Chase Wright: Oh, I was just gonna say, there's a Telegram group for this. Bo, I don't know if you're in it, but I could add you to it if you want me to.

Mercy Boma Naps-Nkari: Okay, so my question is, since Geth doesn't support this yet, how do we, like, do the, testing part? Or that should…

Chase Wright: Well, that's gonna be the last piece, right? So. I know LightClient is in that telegram, and I think there's… one other person from Geth in there. They basically already said that, like. Yes, we will support this, but, the… and Banteg has a PR that's, like, in draft in GoEthereum to implement all this, but until the spec is finalized, and until all the other clients are in agreement, I think that that Geth piece is the last, last piece.

Mercy Boma Naps-Nkari: Okay, I have one last question. So, I was thinking… okay, during, Besu's, last call on Tuesday. Someone said if this needs, an EIP. Is this something that needs an EIP before it gets done, or…

Chase Wright: I don't think he needs any EIP. I mean… A long, long time ago. That… that you… that was, like, the argument anytime we tried to make… changes to some of the APIs, there… there was pushback saying, like, well, it needs an EIP, but I don't… I don't think it needs an EIP. Like, everyone… already implemented, so… I don't know. If someone wants to write an EIP, feel free, but, like… I don't… I don't care.

Mercy Boma Naps-Nkari: Okay, I was just trying to get clarification based on the question that was asked, so…

Chase Wright: There are lots of APIs that exist that do not have EIPs, so I don't think that's a blocker.

Mercy Boma Naps-Nkari: Okay. Thank you. So I don't see any other thing on the agenda we can probably discuss on, unless anyone or you have anything else to point out.

Lightclient (Matt Garnett): I have a question on the tracing stuff. I'm just trying to figure out, like, where… where are we at with the tests? Like, that was, to me, a… A big part of getting the actual spec merged was that we had tests in execution APIs for the tracing, and I think that also depends on this, like, validation. Test format that Felix was working on. That get merged, or are we still waiting for it?

Chase Wright: Okay, so two parts. The first part is there are lots and lots of tests in Banteg's repo. They are not in execution APIs. Yeah, but…

Lightclient (Matt Garnett): We don't really care that much about the… paths that are in his repo. We just want them in the Execution APIs repo.

Chase Wright: Okay, so the only way that they get into execution API repos currently is if the Go Ethereum library can Generate the tests. So, I don't know if you want tests before Geth Implements? But Banteg.

Lightclient (Matt Garnett): This is one of the spec that is tested. That the clients agree on.

Chase Wright: Yeah, okay. And I'm telling… those tests… are currently being ran outside of execution APIs. In order for those tests to come into existence.

Lightclient (Matt Garnett): of them, right?

Chase Wright: Yes. They need to be implemented in Go Ethereum currently.

Lightclient (Matt Garnett): I mean, you said it is implemented in Go Ethereum, so why not generate the vector against that?

Chase Wright: Yes. He has he has a he has a draft PR in in Geth, but he hasn't done it because not all the clients are in 100% agreement yet. So that is the next step. Once all of those clients are in agreement. Then he will open, or he will bump the… Go Ethereum, reference to point to his fork branch slash PR, and he will generate the fixtures.

Lightclient (Matt Garnett): But even the fixtures, like, we also need this validation script PR, right? Like, as far as I know…

Chase Wright: Yeah, yeah, yeah, sorry.

Lightclient (Matt Garnett): The reason that we couldn't do these… Fixtures before was because we couldn't actually… Consistently test them.

Chase Wright: So, that was… I mean, technically, probably yes, but that, the validation script came up because of the call tracer. And… there is a PR from Felix for the validation scripts, and I have left him some feedback on it. There's two of them. One is in execution APIs, and one is in Hive. They are both, like, 99% done. But, there are a couple… small issues that I asked him to fix. And if he fixes those, then we can probably… Merge both of those, and… Call it… call it good.

Lightclient (Matt Garnett): Okay, cool.

Mercy Boma Naps-Nkari: Okay, so. Do we have any other thing else we want to discuss? But, okay, I have another question. What about, Nimbus EL. But most of the PRs doesn't, like, really touch them, or the… Is this something we should also let them know?

Chase Wright: Are you saying that Nimbus EL is not tested in… RPC.

Mercy Boma Naps-Nkari: No, I mean, in this whole conversation, like, it seems like they're not part of it or something.

Chase Wright: Oh. Yeah, actually, that's a good question. Are they in… Is Banteg even tracking them? Yeah, it doesn't look like it. I mean… No. Yeah, I can bring that up. I mean Nimbus EL fails… Like. 70% of the tests currently, I think.

Mercy Boma Naps-Nkari: Okay, so… I think that's the only thing we have on the agenda. Do you have anything else? I guess not. Okay, thank you all for joining, have a nice weekend, and see you all next two weeks.

Transcript adapted from ethereum/pm, CC BY-SA 3.0.