All Core Devs - Execution #61

Transcript

[Music] [Music] [Music] you

you okay we should be live so welcome to meeting 61 first item on the agenda is reviewing previous decisions made in action items so reviewing meeting 60 notes I guess I'll just quickly breeze through this the decisions made decision 60.1 go with a six month time frame for hard Forks for now that was discussed on the last call any comments on that sixty point to a movie IP 10:57 into the

accepted category 10:57 is prague powell that was discussed in the last call decision 60.3 remove VIP six five five four now as it's a superset of 1829 that was the IDI 255 one nine pre-compile and I believe 1829 is the is REM COEs general elliptic curve linear combination VIP actions required did have folks been reading through all these at the start of each of these meetings I don't want to waste time on this if that hasn't been the case previously with zero discussion so I think is to review them okay I'll just

keep going then we just have a few of these action 58 point one cat herders to look at updating a IP one status work in progress huija do we have any new news on that I have a PR okay so I guess yeah great if anyone wants to give feedback on it you can Tim feel free to throw that URL for that PR here into zoom or into into the Gator okay great thank you sixty point one review time frame for hard Forks in June to refresh memories hard Forks in June what does that mean exactly

it means that in June ah sorry we should write okay we examine timing Thank You Boris my English parsing is not up to scratch this morning I need more coffee as well that makes sense so it's not June yet so I think we can keep going sixty point two daño to add nine month out hard fork kick off two timeframes Dano is gonna win the call uh yeah I'll be done in about five minutes excellent thank you and feel free to also well I guess I can post the link for that that's an ethereal magicians thread right okay sixty point three six one five decision discussion at next meeting so this is Greg Brooklyn Pavel

and Christian subroutines and Static jumps I think that is on the agenda for today's meeting already Boris is that accurate don't think so do we want to add that to do we want to discuss that briefly right now or should we add that as a separate bullet point for this meeting I'm here if we want to I'll add it to the agenda for this meeting let's keep moving through the action items glad to hear that Greg and Boris you're both here to discuss it sixty point four down Oh to add list of conditions for implementation and push

request EAP I think it's pull request EAP 10:57 into the hard fork meta 1679 so the IP 10:57 is prog pal is that now down Oh in 1679 just checking yeah and I put on there to say that it is the security audit above and beyond the standard security audit requirements for the typically IP and that it should be considered before inclusion I didn't listen the exclusive reasons because we don't know what the audits gonna come back and there's gonna be of course of discussion on what constitutes clean on it and what's not a clean on it so that's basically a discussion we need to have when the audit comes back okay I see that here under proposed di keys

under 1679 which is indeed the Istanbul meta I posted the link for that and get er as well and and there's that caveat that you mentioned a moment ago down oh that's very helpful thank you should I raise my hand before I talk I don't know sorry we haven't we haven't done that in the past Rick feel free to jump in okay so yeah

Rick Dudley: I mean in addition to the the code audit are is there discussions around you know simulations and all this other stuff and you know test net deployments and all of that stuff or with profile as well is that all like in the same are we thinking about all those things at the same time

Boris Mann: I think that's a separate item in the sense that I just added to the to the comments for this

thing is that there is no point of contact for testing test net coordination or anything other than dimitris retest death work so I am proposing that there needs to be at least one human who coordinates and cares about all of those things yeah I agree with that let's cover it when we get

to that me agenda so Lane can you just add the agenda testing test and the path to that for is simple yes I'm updating the agenda right now to add

six one five as well as that the IP six one five and you said testing test nets and the road the the road to Istanbul I like that that sounds very poetic okay great continuing to move through previous action items sixty point five Martin's Wendy to confirm that a IP 1884 has been merged into the hard floor cometa 1884 is ens interface discovery do we have Martin no no no repricing hope codes okay so there's oh it's 1844 the link is broken okay so it's 1884 I'll fix that

link it's there and it is in the meta okay great okay sixty point six marks by exhausted to confirm whether EAP six eight nine needs to be implemented six eight nine is address collision of contract addresses yeah so my take is we can skip it because 684 is already implemented they're basically the same thing execute I hadn't had the time to to look at it so I think he mentioned that 302 is in conflict with it no not in conflict

I mentioned that create to necessitate that well long time ago we decided on e6 84 I think 684 and that's roughly the same thing as 689 and so I don't think we need to 689 wasn't it do we need to discuss this further should i add it to the agenda or are we comfortable with your determination just now Martin so it's 684 but it's not published as an episode an old-style each you so it's not yeah it's not from August 11 2017

Alex Beregszaszi: I would suggest we take it offline

but the end result should be that a bit aside that six eight nine is not useful anymore than it should be from draft to reject it okay I'm gonna move on to the last one here if you guys are comfortable with that so six deepen the call yeah it's still here but uh yeah I don't have any commentary eat now get something for the next call okay great

we'll leave that as an action item for this call as well thank you just to to phrase it Phil I think the very the question is is the parody of the REM client team comfortable with using libraries for shared pre compiles or would they only want to have native implementations in there in their core client and ping me if you have questions about that I think I said it correctly okay well Google lol Thank You Phil Thank You Boris very helpful cool all right that concludes actions from the

last call so next on the agenda is roadmap do we want to review I guess we should because we're coming up now very close to the first deadline we have a hard deadline one week from today may 17 2015 1 is not really it

Alexey Akhunov: shouldn't really be there but it's only my personal opinion because as I said many times before that just putting something as a

placeholder there doesn't really help too much the EAP is ready when it's ready so I don't know if there's if I just write a bunch of pieces and put it there and and that will basically waste a lot of time to

be in a discussion in court of course like people will try to go through them every single time and then just you know but if it's not specified like I think it's just a bit of a waste of time but that's my personal view on this

Danno Ferrin: so I would like to defend the notion of consideration deadlines one of these removing worries which is different is the train based model of releases where your EAP is ready when he's ready gets

on the next train and those translate the station every six months and if we're gonna do that we need to have some sort of a roadmap that says you really need to be ready by this point or you're gonna miss the train and that's really what this is it's not strictly a if you missed this by one minute then you're gonna get 10% off your grade you're gonna fail the final exam time deadline but it is something that people should be working towards if they're trying to make this trade to try and make it ready or they need to make a gut check call that this VIPs not gonna be ready for the release of five months

Alexey Akhunov: well I understand this the the intention is clear to me but I do not think that AIP is the correct thing to do to be putting in because as far as I

understand the AIP is the specification of the change and the specification of the change from my point of view cannot be produced like you know in a high quality

until Alexa the work has been

Boris Mann: let me just cut you off there so I understand that you have a particular hang-up about any IP until it's perfect so in part this is signaling to understand and plan I personally if you put an e ba that says Lexi will fill in later with high quality but I intend to get it into Istanbul that would be awesome from a signaling is that ideal if like 30 different people all put in things that basically said the same

thing not useful but it's what we've got today and I don't think I think that if you want to bring in a neat past the deadline it will be considered that's that's what a see he can do but in the meantime we'd like to push anyone else not named Alexi get there EEP sit is that a

higher-paying net

Alexey Akhunov: well yes but I think it's yeah maybe I'm just too hung up on this but the we will probably need to redefine what we really

want to do we want the IPS or just we want like a couple of lines of saying that this is what I think should be happening

yeah so

Rick Dudley: I think we're talking about three different things that are currently being addressed with one document and we need to decide how many documents we want to make to to address these three things one is a quality specification which I think Alexa T wants and is something I would appreciate the IPS do not provide that today the IPS as they exist today of sort of a high level proposal which is fine but then there's also this third thing that has come up which is the intention to get your change into the next hardboard and so those are three different things I don't I don't want to blow up the complexity but I think that

we at least need the intention of inclusion as a statement it's made somewhere and we need a formal spec made somewhere and we need these other things in between yes so yes I'm open to you know changing the names of things or making new things or whatever but those three parts of the process still need to be covered whether we cover them in two documents or three documents we probably shouldn't be covering them in one document that gets revised three times I think that's a little confusing

Greg Colvin: the purpose of the last call process was to get a fairly high level of quality in the IPS obviously as

clients implement problems will be found and changes be made but the purpose of that process is to say this thing looks pretty good and let's go and I don't I don't right now we don't have that requirement probably a good idea

that commute if you asked me a question I couldn't hear you was that just a some background noise okay any

Lane Rettig: any additional comments on

roadmap or the possibility for multiple IP tracks I know Aleksei posted a comment here Alexa do you want to expand on that I'm sorry that was Alex do you want to expand in your comment here Alex about not having

different categories yeah sure I which is reflection to you - a boy said that

Alex Beregszaszi: I don't think it sends a nice signal to that side there are different categories of people trying to propose stuff even if it was made as a joke I think we just have to ensure that people don't mistake it

I'm

Boris Mann: prevented from joking trick I hear you Alex so I personally would prefer if Alexei could signal which eats or categories of heaps he thinks he can get ready for Istanbul I think that would be a super useful signal and I don't care if he uses a smoke signal but it would be useful unless we're gonna change our process in the next seven days

Alexey Akhunov: yes so I can do that I can describe we're broth ruff you what what I would

like to get prepared but I can't write the IP because from my understanding of a peat has to be a formal specification so if we were agree that the IP is basically not a formal specification I'm going to do this or tell me what other things I need to produce

so Alex you

Boris Mann: you and I did updates two two three three what if we say eat xxx alexei to be d but here's the name of it so then at least we have a record in in the same spot so in 1679 we would actually have that listed

Alex Beregszaszi: my

understanding of the process was that the first deadline lists VIPs which are in draft mode and we don't really have a concrete specification what draft mode means even for quarry IPS but I would assume the draft mode still means that it may not be fully specified it has a general idea and it is likely to to improve the process so probably the draft mode doesn't mean that it is fully formally specified yet but it has the intention that it will be and with the first deadline my assumption was that

whatever is proposed by the first deadline is the list of VIPs which are going to be deliberated on and with very few exceptions new VIPs should not be possible to be proposed because that may just lengthen

out the process in fact they can start being proposed for the next part of work yeah I would say I'll suggest not to be very kind of I mean

Alexey Akhunov: I understand that people want to have these very foreign procedures but I would like to ask what is the purpose of of these of having this list defined before the 17th of May is it this list that we're going to be keep discussing

for the next like save four months is that is that the purpose

well it does I mean it does help if there is a list where we know ok so these are the things that I should focus on and look into and try to floss with and so in working groups or whatever to weed out the details even if it isn't finished it's it's a list of you know good things that if I want to help out and dive in these are the things that I should look into ok so that's very good to know thank you and maybe it's

Tim Beiko: worth highlighting on the

client side it kind of puts an upper bound on the changes we have to plan for like there won't be more than all of these feats that make it into the art fork and that's valuable to know

yeah just to

Rick Dudley: just to echo that that's the whole point right I mean we need to figure out what code is going to go I mean sighs I'm sorry it sounds pedantic but we have to figure out what code actually goes into the fork and we need to know that sometime in advance and if we just say ok here is the gate and you know if you miss the gate you're not getting into this forbidden to the next fork I think that simplifies

everything overall they'll be generally very good

to have his deadline and at the same time the only the only comment I had before was Aris worried why Alex Hayes changes were not included knowing that he was planning for a free heart Forks to deliver one point X and it would be enough if Alexi just includes day there's some suggestion on which of the IPS from his list will be there so 17th would be great to have to see the upper bound as it was mentioned before okay so I can make the very rough list of things that I would like to be included but there won't be a a IPS essentially in the in my understanding so that's fine so I think

Boris Mann: I suggested it's okay if it's not

yipeee ease but Alexei I think Alex is correct is that likely what would be good is if you created draft eeap's that indicate what those are and then those can be PR din 21679 so draft e means that it's basically filled out you can have a bunch of sections that are TBD but we have a number assigned and we know the name of it and basically a rough abstract that says this is this one so we know what we're talking

okay it sounds like we have rough

consensus on a couple of things so obviously there's still a little bit of a question here around what exactly is an EIP and exactly how well form does it have to be by which deadline but to try

Lane Rettig: to kind of tie a bow on this it sounds like we are agreeing that there is value in these deadlines number one number two that this first deadline should serve as an upper bound right so the idea here is like this is the maximum number of changes that might go into a subsequent hard fork and to Istanbul in this case but of course not everything that is proposed by that deadline will make it necessarily into the hard fork and furthermore that as of this particular deadline seven days from today these EIP x' may still be in draft form they're not expected necessarily to be

thorough and complete and of course they're not expected to have implementations ready because there's two months following that until July I believe for that to happen anyone opposed to those that summary otherwise we can keep moving

I think

Greg Colvin: I'd like to volunteer but I'm too busy as an EA IP editor alex is an EIP editor can someone volunteer to just help very busy Alexei with the mechanics of just that di what an EIP is is actually very well defined and as someone just said get them going

get an abstract get them in so we have numbers to refer to for what Alexi wants to do is it's so important that it gets done it helps a lot to have numbers to point at and say these things

Boris Mann: I'm fine to help you with the mechanics of it I think on the other side we just need people or poking people in a CD who can merge them

Alexey Akhunov: it's okay I think we my kind of resistance was mostly about the the foreignness of the IP because my

previous assumption was that the IP is supposed to be a formal spec and then I wasn't prepared to provide this more formal spec so if that is relaxed and I'm fine with that

alexia are you half of happy with the following statement and a IP draft is not a formal spec but at the time when it's merged slash accepted it must be a formal spec because that helps square the circle that I know I'm

Alexey Akhunov: not sure I didn't have time to think about it yet yeah and we need to think about that

I guess the goal the DI piece is that it gonna be a very high quality specification for each other changes but I don't think that can be said about all the pasty IPS which have been merged there's a big variance between quality

across them

Alexey Akhunov: yes I know but this is where basically I want I wanted to get out of this broken window kind of thing is where okay we had the body IPS before and therefore we should keep doing this so my idea was that let's start doing things in a good way but obviously the lots of the process is still going into the sort of old way and so if we decide that we are going to relax this for now then

yes I'm okay with that no actually the IP processes has been made solid by by we editors for

Greg Colvin: for a while now and to just put a draft in the system it only needs to meet the form of

the draft then there's a long process of getting the draft into the the formal specification quality that you mentioned if we just need something at the beginning of the pipeline yeah the very beginning it only has to meet the form there's a program which will merge it automatically if there's no problem

Lane Rettig: I think we're probably on the same page with respect to draft and alexey has offered to to create those placeholder EAP is boris has kindly offered to help him with that process let's let's keep moving

thank you Greg

Alex Beregszaszi: can I add one final addition

I think the list of proposed VIPs kind of just say the scope of changes but they not don't necessarily have to find a list of EIP numbers and but by that I mean maybe some of the even current proposals maybe will be split into multiple IPs and the same could apply to stay translated yeah peace maybe it is enough to have a single state state rent related AIP merged by the 17th which defines the scope of changes and maybe throughout the process of defining those properly in the IPS we could end up with multiple IPS and I think that should be

fine as long as we have at least one which defines the scope

roughly Alex are you volunteering to huddle with Alexi later and help them to just get this going actually I think ok so I didn't have a problem of actually writing those things I'm alright thanks folks and and Alex thank you for adding that commentary this is clearly the first time we've done this with with a hard fork so let's kind of see how it goes and I think there's definitely scope to learn from this process and figure out you know exactly what role the eye keys are and whether they can be split etc

you know what the scope needs to be the next item on the agenda is VIPs do we want to review the list of the IP is currently scheduled for Istanbul are there any more a IPS to add I'm just browsing through the the comments here and the agenda to see if I missed any so I believe the canonical list of a IPS is in VIP 1679 is that correct scheduled for Istanbul

Boris Mann: yeah so if you want to get a view basically so six one five is still something that we'd like

to get in so those are the ones that are already fully in Wiki link there that Trenton has just posted have a have an extended list of other ones that are people are like yes we're gonna get this ready and PR them in as well and transmitted and I have been teaming up on keeping the the wiki up-to-date so that we can get even more of a view on the Train of other eeap's that may still be proposed

excellent to be put in there is some debate as to how the iake 1344 and one I just mentioned

should be implemented and the author of that VIP 1965 is trying to get that in is trying to get the attention of people on ethereal magicians so there are there is some contention on some of the ideas in this right now thank you that is just to state that is that

Boris Mann: there's contention with thirteen forty-four and essentially an opposing

even yes and that's not listed on the list in nineteen in 1679 right now because the first country hasn't emerged good to know okay are

there any I don't see any in the comments here for for the agenda for today

Lane Rettig: are there any additional EAP use either already on the agenda or not on me sorry already on in 1679 or not in 1679 that that people want to discuss right now Blake to be was brought up actually my understanding of the status of Blake to be is Virgil reached out actually Hudson reached out to Zuko and there's definitely you know Zeke ash would obviously like to see this as well but they're looking for a champion and Virgil has also offered some funding and get coin has offered some funding but unless anyone has heard otherwise I'm not aware that this is currently being worked on

and therefore is not going to make it into Istanbul unless that changes in the next seven days

Alexey Akhunov: so there's another kind of VIP which is not listed but it's there's a working group going on this one is the generalized elliptic curve precompile I think and I think I want to make sure that I will make sure that it will be included so essentially that's the generalization of the previous one

we just won 1889 or something like this I said 1829 I just posted billing Alex see that is waiting on Graham code to finish

his PR which proposes it for 1679 oh so he is gonna do that well he did a PR but it it is out of sync and he wasn't responding okay okay we can attention history so I think right yeah and then also that is also extended now to the because there was a to do in this IP about the the modulus so the original IP was saying that the modules would be less than 2 to the power 256 which means that some of the larger modulus would not be supported most important most importantly will be a less 12 3 8 1 which is used in a beacon chain and but

the now there's work on the even further extended variant which actually will have a modulus up to just the power 15 or 512 or even like 1024 so this is probably going to be extended in the IP or something like that

Alex Beregszaszi: as a second note I think there are probably over five maybe after 10 draftee IPS which haven't been merged yet they are still a simple request and seeing that people want to propose them as part of the Hartford so maybe the

editors have to to help a bit more or maybe maybe we have to allow merging the the proposed list without the traffic being merged but I'm not sure what's the best approach because the the I think we have some predefined rules what quality like basic quality and a IP should have at a time it is merge as a draft before that it only exists as a pro request and some of these proposals may not be up to the quality requirements of being merges a traffic and we have set that only those

which are merged as a draft and show up on the EIP study theorem or website only those can make it into the 1679 list so probably this is a question we need to answer whether this is a good approach or we should change it

Alexey Akhunov: well but this is what already starts kind of coming back to what I was saying is that what is the quality so I we just decided that it needs to be formatted properly and that's it and now we're talking about the quality

yeah even

Alex Beregszaszi: even the formatting is often in some of those

Danno Ferrin: just because it makes this deadline doesn't mean it's gonna be included as well this is the first gateway if you

don't make this you're gonna be on the next boat so we still need a that make sure

and

Alex Beregszaszi: the next big challenge will be I kind of expect that we will end up with probably a list of like 15 I piece proposed by the looks of it by end of next week and the big challenge will be had to actually discuss those 15 and had to reduce that set of 15 into a more manageable set

well

Alexey Akhunov: this is why I was asking what is the purpose of having those lists so it probably it's not it's not the purpose of this list to keep discussing all these 15 things every single time only when something is

really actually happened and you should be discussing because wise we're going to be spending the entire time I'm just going through the lists and going to the same list it will be super boring

as well so yes it so that in like we've said the the gate thing a couple of times so

Boris Mann: the next thing that people need to do is get things into an accepted state so they need to sign up to an all core devs say I would like to review eat xxx and propose it for acceptance that is the next step if people don't get it through acceptance and all of this also essentially relies on champions to keep pushing it so we're actively trying to weed out the things

that people aren't going to work on it's not arbitrary

you thanks for us okay that's that's that's that's great I think we're we've we've talked this topic to death it's obviously a very important topic and I think that we should continue conversation on the Fellowship of etherion magician's forum and maybe on future coordinates calls as well let's keep moving so the next agenda item is alexey x' blog post which I think came out just a day or two ago on changing processes in eath 1x alexei

Lane Rettig: would you like to just

briefly share this

i'm gonna step away from the keyboard for about 60 seconds while you do that okay so I think I'm just going I'm not gonna read it out for you because you can just read it's an agenda there's a link to the blog post and so it

Alexey Akhunov: I wasn't actually planning to write it but then I realized that what I what was happening in my head about like what the what I think the theory Magna when X process should be was just in my head and other people think otherwise and I realized that after listening to one of these but the code is hub community calls when they decided or proposed I would say that to drop the 1x and next name and and then the argumentation was at all that just it's

just a list of our items on the roadmap and why should it be having this name so it's just aetherium but to me it was more than just a list of items it's actually the attempt to change the process and I described it in a blog post so you can read it the biggest thing is that it's not something that I premeditated like from the start it's actually something which sort of organically fell fell out of what we are trying to do so for example from the work and tried to organize the working groups around the certain changes and then the next question was where's like who is gonna pay for this work and then another question was like I realized that a lot of people who were coming and trying to do the work they're already doing something else and we have

a very limited pool of people who actually think they can work on stuff and so then I started to think well if they're gonna work in parallel like how they going to get things in because we have a very we'd have a few bottlenecks in the current process and for example the testing is a huge bottleneck as we repeated many times so this bottleneck needs to be opened up we also have a some sort of a bottleneck that in up in in in the past that most of the prototyping was actually done by the by the three usually one of the three implementation teams and that's also a bottleneck because they were historically quite

busy with the optimization of the clients remember between the Byzantium Constantinople so it was kind of unrealistic to expect them to also do all like a tons of implement tons of VIPs so we want to remove I mean I think we want to remove these it is these two bottlenecks so the ones which is the bottleneck one day when the client with indentation teams in the bottleneck on the testing so therefore basically just push this idea of the working groups working in parallel getting the the the proposals to a higher quality and and basically making the process also much more inclusive by seda basically instead of waiting for somebody to pick up a

change just like if you can form the working group to start working on it and then once you get some good quality stuff then you can just push it in yeah so I'm haven't covered like test nets and stuff like this because it it also requires a lot of description and I also didn't even go into ap process because for me is still the EIP process is a it's probably like something rate with if you want to kind of really confuse somebody you just basically start talking about the IP process so that's why I wanted to avoid most of that and I just showed to people from my point of view how these things actually work and not how they work on paper so yeah so I propose that if people would I

just I'm interested to see what you think about this and whether it's realistic or not and I wanted to make it happen but I

I understand that it might not and thanks Alex a go-ahead bar on so Alexi

Boris Mann: I think it's great you we discussed a bunch of this stuff live in in Berlin and I think it's great to have this to point to and be documented I think the eat hub call one of the things that was happening there is that there's a group of people there that

want to help tell the story of aetherium and I think that's where the context of some of this stuff comes from I'm happy to basically have other people help tell the story of what's going on and hopefully people can look at the blog post that you've written and take that in and help us as a community tell the story better the one point that I will make is that if you haven't already there's the concept of the Osborne effect where basically it was a PC in England where they had a current version that was selling really well and then they announced a new version and sales on the current version totally dropped

off and killed the company and that is relevant context for exactly what we're working on here I don't know how to fix it other than I think part of it is absolutely that we need to improve our process but the other half that I'm just gonna like to basically call for help for again is please help us tell the story of how a theory is evolving

and so

Alexey Akhunov: Thank You Boris that and actually another thing I wanted to say that when we were at Stanford in January so one of the things I was trying to propose which I don't think kind of worked is that essentially I was trying to propose the

monopoly on the changes in the etherium but so essentially say okay so we have we're going to only just do like really really critical changes and everything else basically has to just go away and because we're just trying to fix the change so it doesn't really break break down and I realized that you know this doesn't doesn't really work people want more change so once you allow one change you people want more changes and so then naturally it started to shape up into the process which will basically just increase the bandwidth for all kind of changes that people want to put in and so that's why I think there were some comments about from Frederick at who was also was in in

the in Stanford he was saying oh the actual meaning or if you one acts really changed since the beginning of the year and a lot of this change was happening in like in a couple of people's heads not in like we weren't really talking about these kind of changes so I hope that now it's a bit more in the open so it was my kind of hubris to try to actually presume that everybody is following me on this mental journey but actually I realize no no no this is not like everybody's still thinking about what we said in January or what we said in December and we they haven't really moved on so yeah so it was my kind of my mistake that I assumed that everybody is moving

at the same speed

Boris Mann: yes and we have a ton of other stuff to do like testing if we expect any of this to get in

thanks for next say for folks who haven't read the lectures most recent post please check it out I think it does tell an important part of that story and and I feel like I say this after every agenda item but like we have a lot more to discuss so let's continue that conversation in other places

Lane Rettig: to keep moving through the agenda I did just make a couple of updates to please refresh if you're looking at it just things that had been missed previously in the comments the next item is working group updates Aleksei or anyone else do you have updates

to share on the heath Linux working groups yes I think I will give you the update on the statement working

group and I would let others please do that so one state when basically

Alexey Akhunov: I'm currently still trying to test what we call now the semi stateless client approach which is basically the evolution of the stateless client where whereas each block is accompanied by this block proof which allows to do two to execute the block without any access to the state but now we realize that the the penalty of the block proofs is quite large so an average is going to be about 300 kilobytes for actually one megabyte per block proof per block as it stands now oh and if we just do it further for

the contract storage it will be about 300 kilobytes but this is it's possible to reduce this by essentially making the status client slightly more state stateful so and then you can sort of by by by this figure by setting how stateful you want to be you can actually reduce that block size further and further so my task at the moment is to try to figure out what is the like what is the sweet spot how much stateful we'd want to be and how much of the blob proof is going to reduce and so the the general idea that if you have a net peer-to-peer network like an aquarium and so each tier would would define what they statefulness level would be like it

that means that how many like how many block crews essentially they would in aggregate I would store let's say that somebody decides that they work in a store 1,024 previous block proofs in aggregated format so that means that if something has been accessed within the last 1024 blocks or something has been created within these blocks you don't need the proofs for that anymore because you have seen that already so only give me the new stuff which hasn't been accessed or changed in the these blocks and so that dramatically well hopefully the data will show that that dramatically reduces the the overhead and a bandwidth but also it means that the peers who have a bigger statefulness let's say that you

have a like a real big peer with lots of memory that they would say I would keep 16,000 blocks and they would provide too so let's say that this pure is surrounded by the smaller peers that can only they say I only want to store 256 blocks then the big peer would actually be able to generate the troops for the smaller peers which is going to be larger so this is what I'm trying to trial at a moment so that the data structure which allows you to flexibly to basically generate the proof of any level of statefulness so that appears essentially agree with each other like I'm gonna be this stateful and I'm gonna be that stateful and depending on how the like what is the what is the minimum

statefulness they will exchange different walkthroughs so but this is really tricky and that I'm still debugging through some some edge cases but hopefully that would finish soon and then there then I will put this weak output the first Ti please into the state rent which will essentially only deal with some accounting of the of the of the of the of the state and also they will put the groundwork for introduction of some of the penalties for the state extension or for exte state expansion like creation of the new accounts will become more expensive and the creation of new storage items will become more expensive in the first hartford

according to my plan but the biggest biggest change in a trend a proposal with an aversion for would be that it is very likely that we will not charge the rent proportionally to the size of the contract but simply proportional to let's say it's the code of contract so that would mean that we will not have a vulnerability to the dust Gryphon attack and we will use the state semi stateless client approach to or to sort of to mitigate this

anyway so that was my stuff and if anybody else wants to talk about any other working groups please

Rick Dudley: yeah I'm happy to chime in with the the fee market stuff um

transaction change stuff I forget the name of the working group right now but basically we're still talking with different groups looking for funding we have some devs lined up but we would like more you know we're constructing a budget so you know having a pool of devs available and interested in doing the work would help with that it's a fairly complex change I mean it changes how transactions are priced so obviously I have concerns with that around testing deployment you know test net deployment I think that we should build a reference wallet to show while it developers how

to interact with these new transaction types effectively that obviously needs to also be sort of technical requirement discussions we need to actually nail down exactly which features which which parts of the proposal we're trying to implement exactly there's a there's a lot of it I'd I'm pretty clear to me that it's not really possible to get it into his temple given the scope of the change but yeah I'm like I said we're forming the group now and I'm looking for developers who are interested

[Music] thanks Rick so relevant to you and I

think maybe to other eath 1x working groups as well I'm just curious is there any particular channel where you're communicating about both budgeting as well as need for additional developers is there like a wiki page or something where this is all laid out where people can find it yeah so I'll just

Rick Dudley: I'll just be frank you know in my professional experience we don't uh you know solicit the public for these sorts of things so I I don't have any experience myself nor does anyone in my staff really do that Wow sorry that was probably really loud no one myself nor any one of my staff does that sort of solicitation if someone wants to actually that's not

entirely true we have one person who might be working on that in the future but we're not quite there yet I want to make sure that we have at least one commitment so we're funding before we you know start spending time going out and

thrash in the bushes so to speak to your men a question Lane so Alexi

Boris Mann: has been coordinating a bunch of proposals for the working groups to look at funding and he's sort of quarterbacking that and keeping everyone updated and at least on our case with the EVM evolution we've been feeding Alexi updates and so we're working on

six one five to the level that we can and it tend to merge it we can talk about that in in the next section with that you know light work on coordinating in the sense that the pre compiles the big question related to EVM evolution is there will be certain things including the the Blake pre-compiled that may not need to be a pre-compiled if we can get core clients bbm implementations tuned we know that we either have the people who could do the work or can easily get some people to make time available if funding was available

but currently funding is not available

Thank You Boris thank you Rick thanks alexei for your working group updates any other eath 1x working group updates anyone else want to chime in before we move on I'm putting together the testing working group but I think I'm just cool Thank You Zack all right I'm gonna move to the next agenda item do people want to meet in July Boris I believe this is yours yeah so

Boris Mann: I'm harping on this to try and

do the time lining thing because if we want to meet in July that or late July that is now only kind of ten weeks away and I'd really like to give people at least eight weeks heads up and so just trying to get a basic temperature from the room should I gather together volunteers and plan to meet in July or do people want to get the work done just online

so I have a couple questions one are we expecting a similar group I mean I guess I'm interested in meeting if we

minimize travel distance for everyone and we sort of have a consistent group of people like I rather do it online but if it's like you know the same 30 people that with some you know plus minus or whatever meeting in different parts of the you know again to minimize travel distance or everyone I'd be happy to do that yeah I mean

Boris Mann: my assumption is that if the working groups and others need to coordinate that this may be an opportunity to do that together or people can meet separately you know I'll make sure to add the link actually again to the voting there's the thread in ethereal magicians

so I'd really appreciate if everyone has a look and leaves a comment and just sort of yeah your name's like nah let's just be online that's totally cool then I don't need to keep bringing it up but if people do want to meet we should decide soon did you say it's gonna be in Seattle the two leading places are Seattle and somewhere in Canada

I'm down for Seattle or Canada god bless

Lane Rettig: Thank You Boris I posted the link to the magician's thread in all the channels and all due respect to Canada in New York is currently tied with Canada

for a

second place so please get your votes in if you haven't yet

Greg Colvin: I mentioned it it would be good if people didn't just give a preference but also a possibility you could prefer one but if it's I can go to one that I can't go to the other that helps

yeah we should do some sort of ranked ranking or some things it's not ranking it's like I prefer it's not just that I prefer I cannot go the week of the 29th and other people probably had similar constraints you know things we needed up for this I was gonna

say we probably wanna do quadratic voting maybe we should get democracy earth on it Jesus all right

Lane Rettig: thank you folks Boris thanks for putting that pole together again the link is in all the channels take a look

if you haven't and Greg to your point I'm sure Boris will factor that in do not leave a comment there we're we're a small enough group that I'm sure this can be done manually for now but would be fun to have that built in in the future okay next item on the agenda is an item added here by David Murdoch on from ganache and mentioning aspect compliance issue issue in guest David do

you want to bring this up briefly link it so what this is is a an issue with death or ganache or other other clients maybe we're the vegan RNs one of those we for json-rpc returns or death expects it to be a quantity but the spec says it should be data and so ganache is complying with the spec which causes we're testing against ganache that's using go aetherium to crash there

they're going through him instance so I just wanted to figure out one way or another which which way things should go there spec should be updated or maybe if this issue at go aetherium only go through him like this and I think they try the P or against they go there and I don't think needs to be for the Pope I think I got ass one thing but the V number is generally across most of the clients I think treat it as the quantity because it's so I guess the

Péter Szilágyi: the complexity here is that you

although the speck indeed says that RS R and s are just binary blobs in reality they are actually numbers so all three values are numbers so I mean I don't care honestly we can go with either decision we can fix it

it's just that these used for chain ID from the from the ERP that was defining the chain differentiation and signatures we started using poppy as the chain ID number so that's why it's treated different right yeah and I'm

David Murdoch: fine with any way it goes

it's just work we brought ganache into compliance recently and now there's issues stating that it

doesn't work with your theory almonds so a frustrating situation so yes

Péter Szilágyi: I guess the question is what need to agree on his whether do we consider these numbers as binary blobs or I mean if now if most of the clients accept it as binary blobs then I mean I don't we we can change out the behavior that's not a problem

yeah yeah but this injustice the only one the other way check parity so this brings up a meta question if we're gonna look and

see what other clients do what's the value of the spec yeah we brought in line with the clients so I guess um you could that is why I was asking that instead of looking at the client we should maybe decide what the crack behavior

Péter Szilágyi: should have been instead of what one or the other client is doing so that's I mean that's probably I think we have this debate with Felix awhile back and he was the one who was saying that he doesn't really want to change it because all these three things are numbers and then the question is do you want to challenge the spec or

not

well one thing to keep in mind is it parity may accept the handle the data both ways ganache doesn't care if it's a data for its quantity RPC type well we'll fix it internally but guess is is much more strict on that I don't know parity that's there but aka apostle's law yes yeah so we deliberately are strict on everything because at least based on my experience if if you tend to accept various different things then

eventually you end up in cases in the end of some in a similar case like JavaScript or hey this is a bullier that's converted to array and add 25 to it or something so that you have these weird things if you are too permissive now maybe this isn't that case but I'm sorry you guys I think there's something that I've suggested the agenda and something that came up with the would explain like later but I think there this JSON that sort of the JSON RPC has kind of grown organically as far as I understand and the in the beginning that was no standard and then now it's actually quite hard to do force the

standard open the the claims so what I suggest I mean so I only started thinking about it the if we want to start up to create the proper standard maybe the way to do it is to actually do the non JSON standard and then put Jason as an adapter in it because we also realized that Jason is not really optimal format for everything and in lots of cases it's very suboptimal so I just wanted to start thinking about this but any immediate thoughts and that hmm so regarding that that's actually interesting question if we consider that

we also have our LP encoding how do we are all being code these are SV values because if we the math integers then it might actually make more sense to keep the Jason's back in line with the constants back well I don't know how we encode the them in in ROP but I think they're just numbers which is basically kind of string of bytes but what I basically suggesting that we have essentially kind of half SPECT the with the the network was a protocol which is the east 63 which is basically ROP based and we have the

light client 1 and so what I'm basically thinking is that what why don't we just create the similar one with a similar kind of flavor of commands which could query the data out to the node and that would be the standardized way to get data out like about transactions blocks what-have-you like storage and then we put the essentially the layer of JSON on top of that and so if people don't want the JSON they won't need so it will be like an adapter like a module kind of thing and we keep the standard really tight on the one arrow p1 now I am but that's essentially what you're proposing is to take yet another interface which which is just a huge pain so

Boris Mann: I think

there's a couple of things going on so we're in the midst of having components that need specs spinning out and having their own repos I know the EEA is interested in this Oasis is looking at potentially helping spec json-rpc the other thing that I have to say is that there's many more stakeholders around the JSON RPC layer like wallet providers and other middleware so I think we should encourage those people to form around it and create a spec and work with clients including how to do vendor extensions or extensions as far as I know those efforts are ongoing so that is happening right now and they're going to do PRS

against clients if there's issues that are found and the other thing is this is a layer that doesn't break consensus for the most part and it means there's no reason to tie it too hard Forks so that anyone who wants to work on this can work on it and improve it without any blockers the there's some coordination factors in the sense that if clients randomly change json-rpc at different times than the higher-level middleware gets into a bunch of issues

but that's also something that I think

those stakeholders can figure out I mean I'm in favor I'm

Rick Dudley: in favor of others decide amongst themselves but I would like to get any IP or something I don't know quite what to call it - to unify this back as one of these middleware providers the the difference the difference between the clients is like really frustrating because a lot of it's needless so if we could someone start some sort of project to unify that it'd be appreciated I'm happy to do that for certain things like Pippin Elias but not for everything

Boris Mann: already the next step is ideally just like dev p2p has been spun

out into a separate repo that the JSON RPC spec gets spun out into its into a separate repo and people work on specifying it and the eeep then references certain versions of the

of this back so just one more thought along those lines I kind of like the idea but then one thing that I think the entire ecosystem was missing forever was actually the test case a test suite that can actually that knows implement the proper are received and corely and I know that at a certain point we did a few explorative things on hive

on top of hive but basically didn't really have the manpower to keep maintaining them but I think if we could clean up and maintain a proper test suite for our PC that would be actually that would be the way to do enforce that every client implements a proper RPC and whether if some somebody's wrong then we can fix it immediately instead of because currently it kind of boils down to random people like you guys opening an issue that hey this doesn't confirm to parity and oh yes it doesn't then it's there's just no form around it so I think it would be really really useful to spin a thousand separate read

want to actually put somebody behind it who can create the perfect my sweets cool all right in the interest of time let's keep moving we'll add as an action item to come back to this topic in one or two meetings from now so we'll also include the link to the the EIP that Boris just shared as well as there's a attempt at an RPC spec that Casey put together previously which is yeah V two one seven we'll take it from there thank you all right next on the agenda is a IP six one five this was brought up on the last meeting I'm just going to read what Brooklyn said on the last meeting on this topic so exci six one five is VIP subroutines and static jumps for evm from Greg Brooklyn Pavel and

Chris and on the last call it was said that this was discussed in Berlin some concern was raised specifically about the complexity of a number of the opcodes discussion about perhaps breaking them out into separate pieces still hoping to get this into Istanbul Brooklyn Greg etcetera do you want to speak about this yeah so I think that pretty much so fools I think the authors are pretty happy with how it stands right now and very much looking towards to pushing this forward to Istanbul there's a couple like literally one or two holdouts that's are very unhappy with a number of changes in it

so there's still some discussion about pulling other things out into other AI peas but really fundamentally this work does come as one chunk and I think Gregg may be able to speak to that even more but yeah we're very much at this point happy with how its dads and hoping to push forward for Istanbul thanks Brooke Greg did you want to add anything yeah the the complexity thing keeps

Greg Colvin: coming up and the desire to split it up and backwards compatibility I've got one last idea that I think will probably

take care of the backwards compatibility problem that that could be brought up somewhere else but it's it's pretty much been ready to go for for almost two years now and I think recent discussions have hammered out what problems there were in the spec so I don't see breaking it up as any help it stands together as one piece and when you go to implement it it it really doesn't take very much code and what's the core things are done the little things people want to split out or just not that hard to implement

once you've had the core so I don't see pushing do cut things now just push word on to the solidity and other compiler authors and when they start into that work they'll probably wish we hadn't split those out

cool so just to be explicit

Lane Rettig: I don't currently see this listed under the hard for meta 1679 so are you guys still planning to PR that in there in the next week

Boris Mann: yeah will will PR that in I think we were trying to figure out if we were gonna do a last call in between but we'll just PR it in and make sure that it's officially listed

great alright thank you

Lane Rettig: okay we've got a couple more items left so the next item is one that we added earlier testing test notes and the road to Istanbul Boris I think you suggested this topic did you want to speak to this at all

Boris Mann: I left it in so that Zack could talk about it in that I just keep asking who is pushing this along since I know nothing about testing Zack are you gonna volunteer as tribute

yeah yeah so we're

Zak Cole: working on the same testing initiatives with the EEA I'm leading the testing working group there so doesn't really make sense that we would I'm trying to create a bridge between those two like

defining conformance testing bunch without other stuff so apparently I'm drafting a road map and a list of requirements including like Korean posts aren't for testing so I'd like anybody's coordination and their collaboration on on this from anybody who's interested or has any strong opinions on the matter so I've coordinated with people like Dimitri and others who have been working on testing and we had an initial call and we want to we're gonna set up another call so on the on the last call I talked about hiring a developer to help argument a lot of these and I think I found a pretty good guy so we're gonna push a ball on that we also released our open-source testing framework so check

that out we're incorporating a lot of those tests into it so another thing I need is like I'd like input from each team you know we need to know who identified who the testing lead is going to be so we can all coordinate and I think that we should have a kickoff call like next Thursday at around this time and I'll share the link in the chat the date prior and week the day of so please take me if you'd like to coordinate on all of this stuff and I'll post

a summary and start a working group ring on hearing magicians but yeah that's about it awesome thanks Zack exciting to hear the testing moving forward

okay still the topic of the test nets this the caller we coordinate some of the young named test Nets like Roxton and Gourley and koban or is that out of scope of this call I think that would be appropriate so I've reached out to the Gourley guys I had a call with them two weeks ago so I'm gonna try to get them on the call as well so we could cover that those talks too so in terms of test nuts like another goal where we have at the EA is launch a test net by q4 but I don't think it really makes sense to try to duplicate all these efforts I think that it would be better if we had a bridge between the EA and the the public if they're in community just you know it

doesn't make sense so that's what I'm going to try to navigate for on our behalf just looking at the proposed agenda for example its lists August 14 as our current target date for for upgrading a test net Rob's doing early or an ad hoc test net Nuge Rick couldn't hear you just now what was that oh sorry could Jesus be his internet hates us apparently Rick you want to try one more time or do you want to send a text message here Rick you go a little

bit yeah omi is my connection gotten has gotten really bad I hear you now go ahead sorry about that what are

Rick Dudley: what sort of test nut is EEA looking to deploy

so

Zak Cole: they launch a test not to ensure conformance to the EEA spec but from my understanding the EEA spec is really going to be pretty much the same thing as the public test that spec except they have an additional privacy layer so from the privacy layer down everything is the same it's the same exact system so I think that it would be great if we collaborated on that and

pulled all of our resources rather than just segmenting all this work because if we're working Intel scheme here we're working on testing at the EEA I mean and we're working on testing somewhere else and somewhere else I think we should just have a more coordinated effort so we're not duplicating our work and we're making the most of the time spent as well as what kind of well so personally I think that the biggest benefit from the EEA is that there are old dudes who have been doing this for a really long time and I think that it would be great to understand what their processes are and standardizing a lot of the things that we use currently on the Internet

like for instance Charles Neville is the guy that led a lot of you know standardization for different things I think that those processes and understanding how to approach those from a more enterprise perspective will allow us to create a more coherent and cohesive community and ensure the continued adoption of this technology

or poking together yeah I definitely second that so my next question is when you talk about the privacy layers it's just arm or is it something else it's not just forum it's anybody who would like to participate and conform to the EEA spec I'm sorry yeah okay that was I'm sorry

so does that so right now is that or like what is when you say the privacy layer on top how is that being like what's the actual code difference how is that being realized as code so

Zak Cole: the difference between what quorum does in terms of privacy so quorum just has gas and then they have like an enclave which acts as like somewhat of an intermediary between these two points it's like similar architecture to Pantheon's what Pantheon is doing so Daniel could probably speak more on that

I'm sorry I'm asking this sort of backwards I'm just trying to what is the actual like if you were to build a test net that met

this back versus a normal gas test net I assume there's some additional code yeah so

Zak Cole: the additional code is generally some sort of privacy on plate other than some things that I noticed at the face-to-face meeting or people are talking about adding different op codes and I think that that's really kind of like not a very good thing to do because if they implement some think that's supposed to be specific to the EEA then they're gonna have to fork that and maintain that which is just a huge pain and I think that everything can be done much more effectively and there should be a lot more coordination and collaboration between all these different groups who are working on the exact same

system guys sorry to cut you off Justin we're getting very close to

the 90 minute mark here this is important and interesting but I think we can probably move some of the substance of this conversation to another channel like like the magicians so thank you for sharing that Zach really yeah appreciate your point about more collaboration with some yeah I think that's an interesting question I just want to make sure we save a couple minutes here for testing updates Martin do you have anything to share on testing especially regarding Istanbul Yankees no no that was yeah thanks alexei okay so yeah I've

Alexey Akhunov: been talking to Dmitri for quite a quite a lot for last week and I think he is not available at

the moment but we'll come back so what we're trying to do is that essentially is the part of my blog post I talking about that the testing is definitely a bottleneck and one of the reasons is that it's only like the aleph is the only implementation which currently allows generation of these consensus texts and obviously that's not good you should be able to generate consensus test out of any reference implementation so and so Dimitri has been working on this thing called we tested which is basically - Jason it talks to Jason or PC to the client and it tries to generate those tests and at the moment it does work only on a left and not all the tests work so it's trying to do the

- trying to figure out whether we can do it Syrian for example and I've been working on this and this is where I sort of started stumble upon lots of things about non-standard today non-staff Jason and actually Jason be in a really bad format for the for this anyway because for example if you do it over IPC then you cannot predict how long the messages and I know that goes human deals with that but for example Ali does not and there's lots of stuff and that's why I started to think about not using Jason on the standard layer or something like this but I'm getting through this and so I just wanted to comment to the other bits about potentially bringing some stuff from EA or anything like this

so after kind of observing for a lot of this for a long time I would say that if we want something to be used for testing it's either has to be like a super robust and high quality like let's say that Awas or Google Cloud or it has to be absolutely everything open-source because we have to have the ability for anybody to run this stuff so if it requires some kind of whatever something as a service but it's not very robust and it's not to go it's it's no-go despite how much you want to collaborate if you don't have everything kind of at your disposal it doesn't really work

so that was my comment thanks alexei are there any sort of

action items on the basis of this or just a so yeah there might be more action items so basically at the moment I'm trying to figure out how difficult this is to essentially to integrate retested into the the goalie cerium and it's there some unexpected issues that I'm hoping the grout but once we've done this exercise we probably will spin out some work for other clients because they will get more of an idea like how hard it is so at the moment I think there's kind of not really many action items but we'll see right yeah Mike's retest up and will be looking into it though we haven't done any implementation yet cool good to know

thanks Dino alright that's the full agenda unless there are any client updates or even more research updates that were not included here in the agenda does anyone have any updates they want to share from their end I do have one it's not really client update just a satellite update that I think about one week ago the Rinka business not was upgraded to the Saint Petersburg for essentially the reason it wasn't updated along with Constantinople is because we wrote that Constantinople half a year ago and three months before st.

Petersburg was actually special so now we actually did update rinkeby key back and I did see that a lot of people have not yet updated during cable notes so just a public shout out that you might want to update otherwise you'll notice just hanging there stuck in the network so please update people that's it thank you Peter Oh Lane sorry I had another little quiet little sort of update but it's I'm gonna do it very quickly like in less than a minute think it's okay go for it yeah so essentially this is what I mentioned the agenda so I'm gonna try to prepare like a little workshop for the

for the Def Con in Berlin which will be in the 21st 23rd of August and essentially the idea is to try to teach people how to become core developers to basically to remove this kind of the mantle of Dark Arts and trying to figure out what it does what does it take to become one obviously with some prerequisites it's a super experimental and I probably will be looking at for some help and I think the first workshop will be probably based on Goa cerium and some sort of testing infrastructure and I think mitri will be helping me as well so but I will be like you know I probably will be reaching some of the go same team to help to help me but you know I hope that people like this idea

Lane Rettig: thanks Alexei I like the idea a lot we

definitely need more core devs I'm happy to help

out with that as well yeah feel free to reach out to feeling and I think we can also have a lot with for example spinning of that's not by a puppet or if you need anything extra that's missing I'm sure it can be arranged cool that's very good

Alexey Akhunov: thank you very much

all right

Lane Rettig: thanks everyone I think that's all we've got for today

so enjoy the rest of your Friday's and have a great weekend and see you back here in two weeks have a nice weekend [Music] [Music]

[Music]

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