All Core Devs - Execution #68

Transcript

[Music] [Music] [Music]

[Music] [Music] [Music] [Music] [Applause]

[Music] [Music]

[Music] [Music] [Music]

you [Music] [Music] [Music] you

all right I think we're live hello everyone let's take a look at the agenda we got Istanbul IPS for the first part we have client implementation updates for accepted and tentatively accepted so let's just go through the clients that are on the call and they can kind of just say if they're running into any roadblocks other than the tentatively accepted e IPS that we don't have all the information for and just kind of where they're at so let's start with has anyone from parity here I don't think so

let's start with another mind I just added the tracker to the today that was repeated for agenda and we've just implemented the chain idea so far I was looking at the other ones I think three or four remaining ones the accepted in tentative should be pretty straightforward for reducing all TN 1 to 8 we'll check our implementation the cost change itself is okay but we would like to run some benchmarks and our current implementation and see if we'll have to write any bindings for the libraries as for Blake 2 should be no problem so yeah I'm pretty positive

there will be a question about the proc bow and the AC arithmetic so they probably stumbled - I think that's on the list this might be a bit more work but I'm also quite confident about it ok great guest oh yeah Peter you're muted if you're talking yes oh sorry a touchscreen keyboard appeared above the mute button I couldn't pass it sorry about it yeah so as for gas I

think so regarding accepted the IPS and sensitive VIPs I think during the last chord F call there was a bit of confusion because we kind of split Istanbul into two one of them that we can actually ship within weeks and there were two really big VIPs that would be really nice to have but require some stuff that cannot be done within the two or three weeks time span so I think it probably throughout the call it would have to do always know whether we're talking about Istanbul one or Istanbul two so if we're talking about Istanbul one I mean the one that we are going to

do in what however many weeks on the test nook there essentially we have I think we had four except deities and maybe two times the ones for us most of them are already merged in there are two two IPS that are not yet version one of them was the twenty 2200 that was essentially a Knicks net gas metering yeah II I P and one day was there were two ideas on how to approach it there was the original one I think proposed by way and there was Alex's version and on Thursday we implemented relaxants version inspected out just to

see how that would work because there was no spec for it but realistically I don't think that works so do some problems were kind of discovered and I think everybody agrees that we should go with wise version so the current actual spec of 2200 so that that one we still have to implement but it's fairly trivial and and the other is Lake - we haven't implanted yet because there were some minor back and forth with the EIP so for that well it would be really nice to know whether the eyepiece final details are done I don't think it's hard to implement but it's kind of moving target now yeah so that's that's about it for yeah for Istanbul

one so you have the our tracker so it's mostly done as for Istanbul to essentially proc Pao and easy arithmetic honestly I haven't looked at those because we decided that those will not make it in two weeks thing yeah we don't need to worry about a stable tool to right now I should have said Istanbul one so yeah if that's all I don't see on the Jumbo anywhere if we can on today's call actually move the tentative ones except don't have that discussion sometime yeah so I kind of agree with marking that on today's agenda there was

this there's this item to pick a block number but I mean first we actually need to pick the set of VIPs that's going so it's that might be something better to do on the core devs get her channel once we see how wieldy the EIP is that what go from tentative to accepted are and so we can see it that's very possible to have a and a lot of that's partially coded right yes so one is for replanting the other one rebalancing should be

trivial because everyone already implemented it bombs you know the rebalance in the nut gas meeting stuff is exactly nixed version just has one more clause in it I mean you just need to add one more line to the code and okay and the repricing EIP is just changing the price of the gas price of three operations I think another code just like I dunno the parity mentioned that they didn't think that the 2200 as is would be ready would be able to go they'd be able to go through a security audit by this Istanbul watch understand

what they said under p.m. who is that there were two versions for that there was waivers and version that is ready and we will going with that for since forever and there was a Lexus proposal from last week and that was problematic but I also said that was problematic so there's no point in going with that one so yeah okay so wait wait yeah go ahead and we're fine way is still backing his own yep and the one that's been proposed forever and we are also a model also well it doesn't oppose I mean it's I think it's fine with either so it sounds like parity get a left everyone's on the

or those three or at least on the same page for no one shooting down ways version right yes okay another mine Trinity are y'all good with that no okay great am I missing a client here I felt like I might be I guess aetherium JSS oxic is still working on that one I don't know where

you're at on that or within that team if you're if you're actively working on that if you're just ewaz them I can actually give okay no problem but sounds like we're good to go on that one then so does that that takes out one of the two ones that were tentative that need to be moved to final is it am I correct there yes wonderful there's the one remaining okay and can someone give like an overview of where we're at on that I've been gone for two or three weeks I can't so we are to 1884 so

repricing set about codes I think essentially that AIP was scheduled for discussion along koreff call since for half a year now and every time it was postponed simply because it was a trivial AIP and it was a key our case we didn't get it but yeah I will just cover it in the next quarter Pole no one kind of ever raised points about it except for yeah they're obvious things yeah hey if you do this it might change the page yeah okay so it looks like thirty minutes ago Powell commented on it and kind of gave

a nice overview in the agenda yeah that store from 2300 to 1600 I I couldn't grasp exactly got it who someone did I thought someone did benchmarks for the for the called Adam decrease we're not talking about the Col David decreased though oh wow I am very lost then yes I think that 1894 with all the data that we have from marking on his back is so not controversial that he

can very easily go straight in does anyone else think that there is any problem because it actually increases the prices and decreasing them soon I'm cool with it yeah is there any objection to that it seems like it's straightforward or I'm on the meta and it looks yeah it looks straightforward the only potential risks is someone assumed that they Stephens two point three thousand will be enough for some operations and after the change it will not be enough anymore so maybe some contracts that

will be accepting transfers will not be able to make cost that they were doing before but it's not really sustainable to have the cheap price on such an expensive operation so it's a kind of tough choice between having an unsustainable pricing or potentially break things but make it more sustainable going forward but this problem is not going away we don't have to rebalance it later on unless we manage to do some stage hearing I'm okay with it too just so it'd be great to

maybe to point at the potential problem to people so they can review the contracts and they are aware of the change but more that it can affect them in this particular way just because we understand it can affect them this way and they may not be able right we've changed the price on this twice before yeah I at least when I was doing more stuff with smart contracts and I would talk to people about it there was kind of widely known that it's not gonna you know these gas prices and stuff like that are not gonna stay the same forever depending on if things happen like the

attacks and Shanghai or if there's like more efficient ways and like things we can do to change it to make it better so I think people would be okay with it personally I think that it's probably not a big issue that when you get 2,300 but yeah you may not fail with calls because if you're doing calls you probably I mean those calls can't be state changing in a way but it might be that you want to do log with something from your slots from the storage and

that might fail I think that it's probably more common scenario that together you get a you get a call or either sound or whatever and you don't have sufficient gas to load the log adults a finger will be great to move it to accept it and just see if someone can run the analysis on the contracts on the existing contracts and raise potential concerns if any of those would fail now even if you do that even if we find some what would be a threshold for not doing

it would it just be finding a lot of contracts or and that's the thing for me I don't think there would be an essential really because we can't just hey there's a wall coming up you just keep running towards it I mean we need to actually raise this that's my okay anyone else have opinions they want to discuss about this or opposition to

this it sounds like majority of people are fine just moving this to accepted barring any major security stuff that this shouldn't come up at all ok I think we're good then Tomas were you the one mentioning the analysis of it yes as we have it we as we had it with 1 to 8 3 now 2,200 I think just this this analysis of the

existing contracts stopped us from releasing during Constantinople and followed us to make improvements so maybe the same tools that were used at that time can be run like the analysis of the expectation for the students to be enough article on cost alone I'm not really expecting any significant contracts to to experience that issue whether like what you call significant contract it's a bit of political things yeah that just sounds like what that sounds like is that we would be doing something to then have more discussion

when we're pretty sure that we're not gonna find anything major and that the edict from the beginning of a theory and it's kind of been these things can change in the first place so if you're a contract developer you should have been prepared for that and that we've never repriced this before and not run into any major opposition after the fact I think are all signs that it doesn't necessarily need to be benchmarked in a way that would cause us to use resources for that that's why I was suggesting to raise it with community so we can

potentially expect the community to run something if they're if they're awareness is there that this may happen yeah I think that's a great idea and I think what we can have the cat herders just like do us with like a sweep on social media to announce that this is happening and that that could be effective for letting contracts know ahead of time so if there is a major push we can we can reconsider does that sound good yeah that's what I meant really Oh perfect okay so we're going to accept it with that one any other comments or questions all

right great so we've dealt with the two tentative ones for Istanbul one the next thing on the list which I'm not able to find previous things about as benchmarks for gas repricing II IPs did we already go over there are other ones no so that's that's why I was saying that there's some confusion between Istanbul Istanbul - because last call there was a lot of talk about stuff - at Istanbul - and there are all kinds of gas repricing things so those are all for Istanbul - okay we don't need to necessarily go

over that today since there are more important things if we want to bring it up at the end I think that would be okay or does everyone agree yeah so I would try to close down Istanbul one and then however much time we have we can continue discussing the IP mm-hmm number two conformance testing that's usually in Dimitri's Court writers conformance testing referencing something else that I wasn't aware of I think it's talking about Dimitri's reference tests okay so I talked to Dmitri recently and he had concerns that a lot of the testing the testing

I guess load is on him and that he thinks in the future or even right now that client should be using retest ëthe to create test cases for their a IPS like the EIP champions or the clients implementing I would need to go back and look exactly at what was said and I wish he was here to speak on this maybe he'll be here next meeting but basically that there can be more clients making test cases that would help a lot and that he would be able to be kind of a coordinator of that and also write test cases while while still not having an incredible amount of

things he has to ride on his own yeah I have actually been writing cases using gas as their transaction test which 128th and also tests for a chain of the up code and tests for my own we probably repricing out and so for us and those are positively and I awkward of Gator channel it'd be great if someone some other client can try out the tests and he is well basically so I know if they're correctly generated syntactically and also if they seem to

actually pass the tests we find such act the same I hope to get the test March into the official test repository and also make right now the fillers are on a guest Fork of mine they're not actually really up for up yeah it's not easy to run them if you don't know what you're doing for them and turn it into binary as well where is that anyone can run so there are more options other than the other opportunities we tested

and I wanted to point out there are two resources for everyone who might not be familiar with testing there is the getter channel for etherium test that I know Dimitri and others are pretty active in and then there is also the retest death wiki for creating state tests with retest death I posted both those links in the zoom and I think that that would be helpful for people wanting to write these test cases and hopefully get more participation there did anyone else have comments on this yes my understanding is that the decision was

author's the IP authors for for the champions to make sure testings happenings for their AIB but then with the understanding that since this is kind of the beginning that we need to iron out what that process looks like and I've been working with the Blake to be guys and I think Lisa's on from 2028 there's still is just a little bit of onion clarity of how how they they can go through that process even the ones that are willing to own it themselves so I think that's just a matter of getting Dmitriy and some resources and something I'm happy to help to meet you with get kind of that those pieces together yeah

regarding that so one thing I know it's real pain in the ass to set everything up from scratch and actually generate the test but one thing that the champions can do is to outline clearly these are the things that need to be tested like for 2028 we need to test a transaction with this data it should be this cost you tested with with the exact cost when it has to move on below and see that no to get the collection right and if there are any other quirky things I don't think there are with me but you know just textual description of what should be tested that's a really

good step even if you don't go the whole run testes and write today some stuff that's a really good start so Pantheon we just implemented retouch that support with our most recent release and I've been working on some of these writing some new tests I've been trying to get Blake to be it's been moving target so I'm not ready to finalize the test that I submitted into the published a PR for and I also you know was on a call of Lewis from start wear as well trying to get at 2028 I've been working but I totally agree with Dimitri that we should move towards something like be tested so that all the clients can support and all the clients can work on

it it's that way when the EIP proposers start writing the first reference implementation they have their choice of language of how they want to implement it and it's a lot easier for them then create conformance tests on what they think their first correct implementation is so yeah I totally agree to be test at this is an amazing evolution of the existing test death framework so all right any other comments yeah I just gonna had because I'm so Louis from 2028 I we uh he went say that we we try to come in and help and understand how Kinkos and there was as we discussed there were some confusion on the process

itself and I think would be very helpful if the quartet could lay out the framework for to like reference to like to who need to talk to how we can help and not only for re IP but in general I think would be very helpful for the future maybe have a call we've the core the champions themselves with d3 or whoever is in charge of coordinating via for two runs testing have a hard for coordinator that would be a good point of contact for that yeah I think that would be good for now because we don't have that

Demetri is gonna be the best point of contact for handling retests death and the framework for testing that you're referring to Luiz so on the zoom chat I have the two links the wiki is not very thorough but it'll be somewhat of a starting point and then if you go into the getter channel for tests you can reach out to Dimitri and ask for a one-on-one and I'm sure he'll be happy to do it terrific would do already I'm already on the ether I just had a question for Jeff for for Martin because he replied and he understand what he meant I saw what you publish to run play 28 and I'm not sure how it relates to to retest if oh it

doesn't so there are specific sets of tests which are called transaction tests and they don't the semantics are different from state tests and with 2028 we don't really need state tests for them because we can use transaction tests at least for clients that implement transaction tests yeah so that's what I meant because to validate the transaction we don't need the full state we just need transaction okay so in so to to just in this case I'm going to rather than ever the question to 2028 do we have now anything we do we need to

do something for the testing I would say not okay great so just think so regard we do not to testing we have kind of two things that need to be done so yes somebody Demetri or somebody needs to implement the test cases but I do think that somebody also needs to dream up those test cases and for example that in my opinion currently most of the IPS have an empty slot for test cases the Festina section is kind of empty and I know that I'm Dre I think gumbo from from Elath he made a really nice table for the in that

gas meter in the IP and that is of course not enough to to cross client testing but it is enough for client developed first individually hacked together whatever unit testing their preferred programming language actually tests that it works and I think it that would really be a very important help from from EIP authors to just have test cases even if it's just a table that okay this is the input I don't know it's an input for the idiom or the input for transaction or the input for whatever and this is what we expect to happen because if you have last and that's already a very good baseline check sanity check to make sure

that the implantation seems to work and of course on top of that we can do cross client testing and all kinds of other stuff but it would really really have to have some fairly conclusive test cases in the Eid even if just in table form okay any other questions about that or anything else with this topic suggest yeah I'm glad one more statement sure just if we are looking at the context of EIT 2028 the collator a pricing it would be really nice if if

the EIT itself would have a test where it says that this is the transaction arabe this one should cost and guess and that would already be a very good sanity check because then I could just implement that in Google and I would see whether I am committed to the EIP correctly or not um even to what Martine made me already did I turn the stood from what he just said I think the only thing that we need is this one one example in the ERP description didn't exactly yeah so even though it might

already be created if you have it in the EIP and just do a PR that has that example that'll complete the EIP from a specific like s formal perspective okay so in this case we can make that happen I'm going to coordinate to discuss with Martine offline too to understand exactly the existing reference to vector and implement that in the inner core like in the text itself of the proposal great okay so next we have tests that upgrade and Istanbul next steps so the block number for the September 4th test

met by the way how a September 4th decided since I wasn't here yes I think it's still a bit in the air originally I think the original work time for Istanbul was two days ago for the test night and about - I think - chord F calls ago I flat out state that that is not reasonable because we don't have a final list of VIPs and so I asked people to push it out by I don't know a few weeks but honestly September the 4th is again about two weeks from now and given that we just finalized the

EITS it seems a bit reckless to me to pick a block number now without knowing whether parity will finish implementing or what whichever client will finish implementing the stuff in time so I'm just thinking that with previous heart works after everybody implemented we had literally months of testing period before rolling out to the test nut and now we just came up with the final list of VIPs and we want to fork in two weeks that's that's a big break yeah I agree so in that case let's just scrap that

date and since we won't be able to think of another date I mean I'm assuming we shouldn't think of another date today until we get more information on how quickly the other all the clients are implementing the e IPS that we should at the earliest next meeting decide but yeah just kind of leave it at leave it at that for that for today's meeting anyone have any opposition to that or a different idea or anything really opposition maybe a bit of a note is that if we just say that are people going to wait on to everybody implements it then and we don't care about anytime timelines then we kind of run the risk

of the whole thing a few delays right lot if we know motivation for clients to actually implant so I would still try to somehow figure out the target that okay for the next product the one that will be all the clients should have everything so even though we don't have a block number we should have a more or less a deadline for science to do the thing okay does anyone have a proposed deadline a date for client implementation of the e IPS since we now have a complete list so do you want to the question will the next call that

will be in two weeks we can do what we want since that would make sense if we want to give it two weeks for that that deadline to happen but did you just mention one week or two weeks for people to implement I forgot what you had mentioned I haven't mentioned it that's why I'm asking whether time is whether giving the one-week deadline is doable or whether we should be a bit nicer and I lean toward two weeks so people don't feel rushed personally I think time for

us to feel a bit rushed okay any other opinions yeah if we were gonna say more than one week then we're at serious risk for not hitting our main ever release date I would say and which probably requires external communication if we think we're not gonna hit but I think one week is reasonable from our point of view so the this list of 60 IPS that we have is not really demanding many minor changes like to that's all only connecting some libraries should be

okay let's say one week that sounds good to me and let's see as far as the test net block number and the yeah the test net block number I don't know what are people's opinions sorry would that would that be a good thing to say yeah so if I if we go with one week that nine to implement that I would say that okay let's have an awkward at all next Friday and pick the block number them based on wherever people are or teams are yeah

something like we could do that we could also have a instead of a full call depending on how much we really need to get done besides picking a block number we could just do it over getter let's see yes but do we have enough participation I don't know how many people are generally regularly on Gator how about we pretend with the call and cancel it if we think we can get away with Gator but propose the call now yeah that sounds good to me we'll do a call in one week okay and that call will primarily are number one

agenda item for that call will be deciding the block for the test at fork and we can also deal with some stuff around naming and the second part of this Stan bowl or whatever that might get renamed to and things like that so and we've already agreed to split a stain bowl into two forks it looks like or just create a second fork that has other stuff but not calling it Istanbul - yes I'm not sure whether you got all the memos on this but essentially the reason why we originally proposed the split was

because there were two big EITS that are are really nice one of them is Cracow and the other one is the fancy elliptic curve pre-compiled and the problem with both of these were that they are kind of done they are kind of almost ready to be used but they need some work that that we don't know when it will happen for crop out we need the security order to be completed which will be done whenever and for easy arithmetic there are still some back and forth on the IPS and and then the reason why we chose to do a second Istanbul

hard for us because we really would like to get these in but they definitely want to make Istanbul and but if we say that okay these are got dropped out of Istanbul and then after we do this to heart form we start planning the next one fully publicly whatever whenever then we kind of run the risk of essentially delaying these two for maybe another year or half a year or whatever and then the idea was that after Istanbul is done we should all teams should kind of focus on getting these two pushed through and then during the last quarter hold for some reason 640i bees got pushed into this bucket I guess it's fine ish we can discuss it but the

point yeah Istanbul to was to not make it a public IP where more stuff can be suggested rather to to cherry-pick anything that that was aimed for Istanbul but then make it due to not being fully completed okay thanks that's a great explanation and okay I think I'm fully up to speed now I just posted a link for the Prague Powell audit github page where we put the cat herders put some of the timelines and some of the scope for the audits so if you all want to check that out basically the final audit report should

be delivered by September 11th is the plan from least Authority the hardware audit report will should be around the same time it'll kind of that that one's still a little more up in the air okay so with that and fully understanding the second Istanbul fork it was there any other comments what point should we discussed Martin's on pooled or staged fourth model the end of this meeting or during next meeting let's do it during this if we have the time I kind of just browsed that briefly and it looked really cool the to the other thing that was requested was x''k requesting talking

about the naming and the eighth magicians forum posts that they did for the name naming so I think that's something we couldn't we have enough time to get to both of those I think before the meeting is done so which one should we do first let's do let's do the timing one first from Martin so Martin if you want to just give a brief overview of that and then if everyone goes to the Isle core devs Geter James made a really cool diagram about it it looks like so that's that's kind of cool yeah it's already familiar with it the idea of being - yeah make the process

simpler for everyone so we don't have to rush to get things in right now we get kind of into a race we accept a number of each and we hope to get everything ready by a certain time and my proposal is to change it and we accept the eve don't say that good we think it's yeah if this and yeah everything is fine and then the other people or those who are have an

interest in it pushes through with the PRS they make the testing happen and once they think that they are complete with all of that they tell the court of Channel hey everything is done we have it in these four clients we have the reference tests and the clients pass the test and all the work for those has to at that point is to say okay so let's schedule it or whenever and that's that call it 800 so I guess the difference that you would do now is that currently what we do is we accept the EIP shut you it and then start implementing it and

you would say that no we would need cross-clamped implantations first before even scheduling it into work yes sorry go ahead nod I was just saying wait so we wouldn't be racing against some artificial deadlines but instead that things take the time they take Dana did you have a copy on my internet was going on stable but now I'm back what I like about this model from the way we've been doing it before is the IPS are accepted

and they progresses they're ready versus before they'll be accepted for a particular UI key and if they failed and they're back to the beginning so this way we move it through as they're ready and then when things are ready we pick it up and go I think that is that is the the major change of those steps can have it outside of waiting for a record of calls to talk about them because it's been one of my observations there's just a lot of waiting around

excellent my main concern everything looks great on this I love it but this seems like it would cause more rapid and more like basically more rapid and more hard Forks in general which I feel like would put a lot of people who run a theory and client software like having to rush to keep up with the stuff so yeah that's kind of a concern I don't think this this model actually wants to make Forks more frequent so you wouldn't work for every IP when it's ready rather

essentially all the IPS would progress in the individual into this ready state and then whenever you you pick that okay we want to hard fork in half a year then you just pick the 25 VIPs that are ready at the time it's ready okay instead of picking VIPs and saying as the form contains will contain these e IPS rather we say that the fork will contain everything that's ready yeah so with this model it makes it possible to do exactly what you say awesome to do one hard work forever Eve and I personally think that it's not too bad but that's only an option we can do it

like also the proposal doesn't mandate either way so from a perspective of anonymity nur of client and people wants to propose the IPS I really like the model because it doesn't eat the law you to apply anytime and start working on it and say okay making progress and now I'm in queue to get accepted whereas today it's more like oh there's this deadline and I need to get everything ready for it I think much better mother from Athens incentive for people to off to propose the IPS

it's easier for two to understand understand to when you need to come apply in a prison project that sounds great I'll call ahead Adrienne sorry well the challenges we've seen with the IPS in general is that they don't get a lot of scrutiny until it's you know hard for can about to go live and then all the sudden people pay attention to it I think there's this value in this and maybe we need to make it very clear that that accepted step is that it's the time that no no once we once we reach accept or whatever the particular wording is for that status that it will go into a release at some point it's just a matter of when now and that might be enough to

focus it but that I think is the element we need to watch it don't don't get a glut of the IPS that haven't been looked at that all of a sudden because a hard fall it's kind of starting to be scheduled get a lot of attention and go back into tentative another risk potentially run is that because it's the IP singly a key focus that in these all the tests are focused on just that II I P which means that if we bundle them together at the last minute into the next release then we haven't done any cross testing between the beeps necessarily until right at the moment that we've decide to package them all in so just something we have to think about

as we shift so if we have state tests probably define the fork with whips that we want to roll out then we can always generate state tests sorry block chain tests which will cover them at least implicit implicitly everything but maybe not yeah explicit possible explicit stuff that affects each other okay something that I think we could you notice that currently the EAP is

wouldn't have gone through all of the checks that would have been gun proposed system because they find implementers would have okay testing teams would also have to say hey James you're cutting out a lot you might just type what you want to put okay so James gonna type execute something to say and I believe your mic is working as long as you and mute can

you hear me yes I can a great servant interruption I just wanted to say that the EIP process was supposed to be similar when the last call feature was introduced at the end of the last call for a corey IP it would go into an accepted state or there was an attempt to redeem that to ready but basically that would be the state where it would be considered for implementation in clients and then it could be considered for adoption in a Hartford but so basically the the process should be

already similar but nobody enforced this at all but the issue people face when they propose in the IP and I don't see that this proposal would address that in any form yet instead had to capture the attention of the core devs earlier in the process yeah because Martin's proposal has stage where the core devs bless a proposal aka how the proposer that its support why to work on it but that should be early enough I guess it's not well defined yet what stage that would happen so actually yeah

I haven't asked so I think the problem with eeap's is that not everyone monitors the EEP tapes basically and when there's the last call what everyone is attention because it's it's not really mandatory for us to do that and we have a lot of other things to do for this the first acceptance something I wrote there is that the client developers express that yes if this we will accept a PR for this means that in order to get that passing there needs to be someone from parity and some are from

gas and other clients that actually say yes we we are not going to reject this if it makes it into a PR and if none of these people are present the meeting where we talk about this the the blessing cannot forgive you so that case people will have to make something of a commitment that they yeah are going to accept it so one of the things the enterprise etherium Alliance does during

their regular calls is they'll list out the names of the IPS that are in final call there's no discussion about it it's just as simple here's what we're discussing final call and if we could add that into this meeting like at the top of the meeting at the bottom of the meeting that might help visibility among the authorized for the Yankees the to go and call I guess the difference is that the final call if there's a final call - no but it doesn't I think then it just progresses in this case the final call me is that everyone needs to accept it we give their blessing so in action means that it's all or blocked and yeah that's good so the blessed status

is is much later in the passage and and I anticipated well you mentioned that the blessing has to be given by multiple representatives yeah but the issue I'm seeing and I might be wrong is that a lot of people come up with a IP ideas but they fail to engage with any core devs to get some feedback on that idea and they're not really encouraged to fully develop the IP and maybe do some tests and and any kind of implementation before getting any kind of feedback whether this is a really bad idea or

this is something we're trying to to look at yeah I am solution to that so one thing that the measure essentially is my problem my personal issue with the wholly active process is that it's essentially the whole process is untrackable from my perspective personal perspective maybe it's probably but so what I see is that we have this sound e IP repo we have hundreds of open e IPs some of them have are just ideas some of them are well developed some of them have been even been proposed but not

discussed for two months and my main issue is that for example if I let's say skip to code of codes because I couldn't attend then I will have absolutely no idea of what a IP in what state is so maybe one thing that would really help if we could have some visual representation of of the progress that eg IP is having so that you know that okay there are these four e IDs that were proposed when one of the most proposed one month ago the other one was proposed yesterday but these four e IPs are in a particular state where you kinda should take a look or maybe in a state where Dimitri as a tester should

take a look or etc so if we get if we could have some visual list of that constantly updating that maybe that would help otherwise it's really hard to do no evil Gucci eyepiece to take a look at think the reason that is non-existent is we call draft and then we have this accepted state which would be after last and if you want more steps which will feedback on that you would need to

introduce multiple stages and define and I'm not sure you want to go into such a formal thing so yeah I I'm gonna I'll think about it it's it doesn't make sense discuss it down but I'm trying to figure out something that would work for me and then maybe you guys can give your opinions on it there didn't need to be any edit stand could just check that

certainly certain things that happen without having to formally put it into the was spending most of my time just trying to follow it I think if we create a checklist for the AP champions for what exactly they need to achieve for the excepted States then that checklist can also be used by us as a dashboard for the progress for the clients so if the checklist contains things like defined spec defined tests confirm with Dmitry and testing on availability of the reference test have

two major client implementations to implement the AIP and get the acceptance for the inclusion in the next Hartford and this checklist will be easy to follow for the champions and they'll have the feeling that they know exactly what to do which is great when we see what Louie did 40 20 28 like he followed with all of us to to get the full understanding of what we need for for DP to be introduced the hard work and he followed up on that checklist will be great for for Anders to have the same understanding of the process and then we can use exactly the same thing to just display at least of all yet piece and see how far they are in

achieving this result yeah and this isn't something we have to fully decide on today Oh Alex jab something else Peter I just have one last suggestion that I wonder if it would be possible to maybe have like one or two ip's taken up on awkward' ev's freshy IPs and people would need to read the abstract and make a decision whether it makes any sense to - for the champion to work in the div just to give an early indication to the champion that this is

worth and this could be the very first step for any that sounds reasonable to me personally because if I'm an EIP writer I want some kind of indication which I mean that step one is that indication to an extent that things should go forward but I think elaborating on that a little more would help people understand what a green light actually means so and that might be in Martin's full write-up I'll need to reread it so yeah it sounds like we have more or less broad support for

this with potentially a little bit of tweaking so it sounds like Peter and Alex might discuss this more and Peter you said you were gonna id8 on this a little more process maybe we can we won't solve it but at least I can tell you guys is that so one of the issues that that bothers me it's really hard to follow the discussions and the progress so generally I knew that on Thursday or on on Wednesday I don't know which day I started implementing the the net S store update and I know that I personally posted at least static comments on the

on the github yeah on the EIP for requests and those were quite beefy updates and variations and I went back and forth had ideas and then went back on those ideas and essentially everyone is doing that and obviously that's how the ideas evolved but I think this is kind of a problem because then somebody else comes along and then there's this huge discussion and then you will need to read all the discussions and go understand every single rationale for every people and only essentially unless you read everything you won't know what the discussions ended up being and and I

don't know maybe we should somehow figure out a way so that we can support discussions but after some result was obtained via those discussion somehow move those results back into a more condensed form so that other client implementers don't have to read the entire thread just to figure out what the end result was for example in the nest or this case I think the the result of the whole discussion was that Alex's idea has certain limitations and then it's not the best choice but you kind of need to read everything to understand that and that's an issue I think there is a feature in aetherium magicians

because they're using something what's the form they're using what's the software it's called I can't remember this before but I thought it had discord yeah I thought it had a feature where you could see a summary of like what people have been saying and I don't know if that feature is turned off because I'm looking now and I can't find it but I thought that was a really cool feature if we had that enabled there so I'm gonna ask about that from Jamie who I think is one of the people who runs it and see if that could be part of the solution or alternatively we could have

someone like the heart fort coordinator dedicated to writing post every once in a while that sums up the previous ad post or something like that like this person talked to this person and this was the result of that and so this objection has been dealt with and stuff like that yes that'd be a good idea right the changelog and then essentially just say that okay the last one week or two weeks there are some discussions

maybe a few links here and there and then just summarize it that where are we at current it's just to try to cut through all the crafts and just get to the the point cool yeah that sounds good to me any other comments on this for now a few people are gonna id8 on this more maybe tweak it and we can update the graphic a little more as we do that I'd say it'll be great if people got better at updating the UI please reflect any any agreed changes as well like really quickly sometimes happen sometimes the IP is a long way behind to consensus on what should happen

yep I agree okay next up we have hard fork naming so that's something that axe ik brought up on an ethereal magician's thread and posted a comment in the agenda if people want to read along while he gives an overview of that so go right ahead axe ik oh and you're meted there you go there's actually an older discussion and maybe maybe we should just name the

second heart for something at any symbol party and that's why I went back to this all discussion thread going back to April where there were a couple of different proposals and I propose to use DEFCON names names and I give an explanation I think that's I don't want to okay so what are people thinking of that

those who've read it I think that's an okay idea I'll have to think about it more personally I think it's a good idea I think people will have comments about the actual names the color of the bike shed but yeah I just think it's good enough my takeaway from Iraq you'd and I had a long conversation on this was that it's a good idea so I'm gonna be Debbie Downer on this Dimitri posted EIP during the opening IP call where you recommend it just numbers for the names and I think we need to get a

split between the public name that we do and the private name that we do because writing in the code of writing in the tests you got to keep a large mental map of what the order of the IPS were when we did our when we did the Shanghai attack yet Peas where those fit in you know what the city names were and now we're gonna have remember the order the DEP Kahn's my concern is that it makes the code that we're gonna have to write in testing and in software and then we're gonna have to maintain in our Genesis files a little hard to understand unless you are very deeply invested in knowing what's what and I you know increases that the mental burden and I think public names are great you know give it these names like you know Berlin London Shanghai and all

those other fun names because it makes great conversation in marketing and we're talking about these in public it's given identifiable name but I would propose that when it comes down to the reference tests are your genesis files we just call it eight nine and ten eleven yeah yeah I agree with you I think that kind of separates because I think internally I'm gonna have to have a reference map anyway and it's super awesome if all clients can just if you can configure any client to know that yeah it's Sarah one sort of seriously but I still think we're gonna need to have the public names that's what I thought xyx proposal was about the public names and I'm all for having numbered internally yeah I'm getting

warmed up to the public names being the Devcon cities but how many okay you already expanded that we'll run out of that we won't run out of names anytime soon so that's pretty cool and let's see okay and then the numbering system that Dana was talking about that sounds good is there is there like I've only worked in a true corporate professional software environment not to say this isn't professional but like one that has like version number naming for certain major releases like I've only done that for a couple of years so is there like a precedent for this and non blockchain industries it's called semantic versioning and it's

gonna develop quickly down to a single implementing number the first number increments when there is a breaking backwards change which is the very definition of the hard work the second ones a minor upgrade which might correspond to Bitcoin Softworks and the third one is basically bug fixes but since every upgrade we're doing is gonna be a breaking backwards change we're changing things that are gonna change fundamental rules of the consensus I think the easiest thing to do is just you know start from start from Olympic or homestead and every time that we upgraded a test net with intensive going to made that we incremented by one and that's where the number comes from it's just a simple count of how many times that we try to hard work Assam to answer your question if you think of on

blockchain world and I guess Microsoft Office will be very familiar for all of us and it has number Inc underneath like 11 12 I think the current version is 15 or something so I think the company internally uses it for tracking for simple comparison of versions and they name the versions with years or just any marketing themes outside so one issue that I do see with versions as I know that at least sir contestants as far as I know : typically they applied hearts so they have a few VIPs that they applied

independent of may not and that kind of been problematic because if we say that okay I don't know Istanbul is version 6 I just had a number and then we define version 7 as whatever and then one of the test nuts introduces in between a fork to handle some scenario and then their version 7 will be different from our version 7 and it's debatable that will cater 7 but it's going to get things messy do Bunter uses a year and month as their version so like 12.16 12.04 means it was on 2012 April and that's been pretty

easy to follow yeah yeah but still nuts I'm configuring is stunning that's I guess you could say that I want to apply hearing 19.4 and I know I think that I am you like the value of the number approach I'm just trying to figure out whether there will be some corner case which will backfire so what Coburn could do is usually they would be doing their VIP so it'll be off of a particular main hard work so you could call there is a 6-1 or eat six Cove and one to show that it's based off of this but then it's their own split

off of it the downside is Kovan won't be able to tell where the gaps are where they're skipping forks then where they're speaking reason but we're gonna have to do that anyway because we're skipping Constantinople for most of the main net to go straight to Constantinople fix yeah but I guess for me that you could always say that you're going from version five to version seven so that's that's kind of doable but for example if what happens if for program apparently decides that they're going to ship all the EITS for if your own version seven but they're going to skip one of them so they're going to ship almost all of the IPS then you can actually call it version 7 because it's

not really all of it but you're not going to add it anyway in later point in time either so if the it gets a bit funky and currently the way to do it is that you just say that now this is not the Istanbul for this is the idol of Berlin fork and boom done and then Cohen doesn't have Istanbul for but if you just have version numbers then those will start to clash across networks so that's why I've always proposed with eo-1 etho to etho three so Cohen wants to start their own thing they started : oh five Covino six you know you could just never let your hard work numbers well yeah but if for example if I had to start a private

network then things get funky again so for a private network do I want to apply : o 5 or 0 5 and you that's I don't I do not want to derail this discussion I'm just saying that there's a funky corner case here boys that I say we can take it offline for all Chordettes and Schneider come back to that if we have some specific ideas that sounds good to me I got like oh right oh yeah specifically

so the non numbering one I'm totally cool with the naming of them for cities me personally it's anyone against that I support it it's ok I support it okay let's go ahead with that for would that start with Istanbul to them would that be Berlin yep Berlin was the first the IP 0 or DEFCON 0 yeah so it'll be let me see if I can do my history right Berlin London

Shanghai Cancun Prague and where is it this year Osaka yep you guys are where we ever do another Def Con in London yeah yeah we could do something like that or just have it Devcon brexit okay so what's next on the agenda we only have five

minutes left are there any client updates or anything anyone else wants to talk about before we go into just the previous decisions made just in case there's anything hanging out there okay so the last time here's some of the decisions we made decision 67.1 was the EIP 6:33 1380 1985 2045 and 2046 are tentatively accepted and will require benchmarks we've already talked through that today EW okay yeah so we actually didn't do that we're putting that off until we

finish up Stanwell okay so then decision 67 point to the IP 1352 and 1803 out of his stable due to lack of championing yeah so they'll be out of Istanbul so there'll be a number of potential oh so they're out of all hard Forks okay decision sixty seven point three all the gas cost the IPS are moving into tentatively accepted still would require benchmarking to be considered on a hard fork again we dealt with that today I think and that's Berlin yeah but so

these were all so all the decisions are about Istanbul too and they essentially about these VIP so I created a if you look at the gas issue tracker for Istanbul I have a list of issues for for the second part of Istanbul act as as far as I know all these discussions are talking about the Berlin parts yeah no I was talking about we do have eleven away to produce all tea and pick up our gas code yeah we do have a view to this so the decision all the gas post oops I guess didn't cover that now so I think what this decision wanted to say is that all the AIP is

that word is that relates to gas cost and we're hanging in the air they have been moved to tentatively accepted so they do not want to downgrade already except to the abbeys rather upgrade hanging once but again these are about the burden foreman yep so that doesn't need to be addressed today or maybe even next meeting let's see EW 1559 stays withdrawn his decision sixty seven point four sixty seven point five used retest death to generate

reference test for all tentatively accepted and accepted he IPS we already talked about that and then finally decision sixty seven point six the next all court dev meeting to be last call for something to go from tentatively accepted to accept it to be included in the october fork that was also completed action item was to get a status update from both the death and parody before the next call i think that happened because they commented on the agenda and on pre on different ones so yeah that's all of the action items from before i saw i don't know if anyone has any more information on this but there was the eath 1x finality gadget working

group got restarted by and there's they had their first call earlier and I think that was Alex Stokes doing it I think they either got some funding or someone just step stepped up and does anyone have any information on that I have I haven't followed it yeah I was on the call there is mostly organizing meeting to try to figure out how we want to go forward and what wants to go on with that there was a little bit of discussion about you know whether it's going to be I saw for it where the clients are gonna implement a full be contained client or if there is going to be some sort of RPC mechanism back and forth to the clients or there was a brief discussion about whether there should be some sort of a witness put on the chain or not that's still

controversial because of latency one thing that I think is relevant to all korjev's as they were concerned about the timing of the hard forks in Casey need to do hard Forks for these and I think they were still operating on the nine-month fixed cadence schedule and I let them know that we are going to discuss Martin's cool idea and point it out that what we're doing to prod pal you know that for big important ones like the finality gadget would be big and important that we're going to go ahead and fork when needed so whether or not we stick with the Cade is something big and important can get a fork as needed so I think that's probably one thing that impacts us the most is that the cooling model that will fit the Markham hose is really going to help with these big tentpole type changes

[Music] okay great thanks for the overview that's it does anyone have anything else I think we got through pretty much the entire agenda all right we're gonna see everyone in one week what's the time gonna be James do you do yep and it off the top of your head cuz I'm really bad at figuring out what we 2:00 p.m. UTC 23rd 2 p.m. UTC ok so that's like what we had for the longest time the normal quote normal date and time that we had

before we switched over right yeah and is the plan then two weeks after that for the one after uh I guess it'll depend on what we talked about on the 23rd if there's not a need to have one on the 30th and we won't okay I'll let the count there just to move the next one and then we'll work it out from there okay thank you so much and thanks everyone for joining we will talk to you in a week goodbye [Music]

[Music] [Music] [Music] [Music] [Music]

[Applause] [Music] [Music]