All Core Devs - Execution #60

Transcript

[Music] [Music] [Music]

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

[Music] you [Music] you [Music] all right welcome everybody to the meeting

I'm your host Hudson and thank you so much Tim for running it last time did a great job and let's see have the agenda open and let's go ahead and review the previous decisions and then the roadmap first of all one more thing everyone hear me yeah okay and y'all can hear me on the zoom call right yeah okay perfect all right so the decisions made from last time were to stop the prog pack

harbin vote actually that was made from 58 so anything from 58 we're gonna skip it looks like the actions required work in progress for cat herders updating the IP one the smaller hard fork versus larger hard Forks to be discussed in the Berlin meeting which is an agenda item metallic to format the currently proposed a IP 1559 VIP Vitalik did you happen to know Matt that yeah I've got the people he's working with awesome okay that is that Eric of italics I hear

some static on here in vitalik okay Oh wonderful so we have actually formed the working group for that well not on not just for this IP but basically if we're just making it implemented so Oh perfect and yep that's we're gonna get to that on the agenda - yeah sure yeah perfect all right action item discusses how should continue to be implemented if the technical audit is not funded and the etherion magicians are here the prog how audit is going to be funded meaning like everyone's confirmed their funding

I should be starting this week hope for next week hopefully and when I say that I mean just there's some logistics to work out meaning people have to sign some paperwork and the funds have to be sent so the down payment is going to be the die that has been collected from the fundraising on get coin so Prague powerwall so we talked about in the Berlin meetup so I'm looking forward to hearing about that let's see for roadmap we're gonna be discussing that today to EITS working group updates please add

more-- ip's to 1344 that's fine and then the working group updates which we'll get to Oh parity etherium added quick consensus for full support for Gourley tests net and the beta and then turbo Gotham to do updates on turbo death which we will have today if that wasn't already done last time okay so we've gone through the previous decisions now for roadmap if you click the link it's gonna go to the etherium wiki that roadmap part is actually a separate item I just realized

that I should have let's see that that would be oh actually no sorry Danna you had something on to gauge opinions on post assemble fork timing if you had a link to anything or if you wanted to discuss that today you can go ahead so the link is an ethereal magicians link where I put up a schedule so during the Berlin meetup we had a little room call of what we thought an appropriate cadence would be after Istanbul for subsequent hard Forks mostly room

supported six-month hard Forks there's a few people in the room that supported four-month hard Forks so in that theory a magician's post I put together a little chart of what the timing would look like if we did six month versus four month card Forks and I think we should focus on six months because if we do four months hard Forks we're gonna get an overlap one month we're gonna release stuff on the test net the next month we're going to consider what's in the EE IPS for the next one before we deploy the main that and then we deploy the main net then the next month after that client implementations are done so the risk that I see there is if something goes poorly with the previous hard fork and client developers need to spend time making adjustments to that previous to that previous network

upgrade they're gonna lose time implementing the e IPS for the next for the next upgrade so that would wind up being a full skip if necessary to the next hard fork another four month delay but if we get six months we don't even consider the EIP to the next hard fork until the previous hard fork is complete and I think that helps the trained model of you know having these scheduled ahead of time to know when the next one is coming so we have these last call for eeap's for a network upgrade that if you know when the next one is coming and you know your stuff is not ready it's easier for you to be honest and say look we're not ready this isn't gonna be fine but we have another train coming in six

months so we'll just hop on that train and I have a feeling that will improve the quality of the eeap's and the implementations that we have coming into these network upgrades are there any common it's about that from anybody yes so I just gonna sort of repeat what I also commented on what this Danno said on his he basically really bro pretty much wrote read out what he put it on a magician's and I responded to him that this overlap that you're talking about between the the like test net of the previous work in a IPS of the next work is only a risk if you're assuming that it's always the same people who do the

preparation for the Hartford and actually developing the IPs however however if these people are different and this is actually the model that I would like to see like where that's why we're spawning all these work groups will come to this later that in this model there is no problem of overlapping because we are having basically multiple people working on this it's not like everything is bottle next on the like a very very small group of people so yeah this was my comment and we'll see maybe we should just see whether we can kind of see how first one goes and like whether we can get enough work done to be able to push it to

squeeze it further but I would like to us to actually at least try to create this overlap and see if it works I think we should start with six month at least once before we go to a four month because I mean before we work in you know nine to sixteen month alright for Cadence's so even six might be a stretch but as far as we're lap on if things go well and the test net deploys just fine you're right it's not the same group of people but if things go poorly and we need to bug why we're getting consensus issues on the test net typically it is gonna be the developers deeply involved in debugging the consensus issues there give me the same who is responsible for implementing so that's where the overlap risk comes from

hmm okay that's interesting I like the idea of six-month ones anybody else just briefly to say the timing out loud to get it into into the notes so April 2020 will be the next one of its six months again I pasted the link into the zoom chat here if you want to have a look so realistically this means if there is another core devs meeting in July that would be exactly nine months until the next hard fork so I'm in support of picking up our pace we just need more people and some more support for tuning process so that we can get better at it so I don't think anyone's against Alexie's point of going faster

we just have to practice yeah I mean if this first one goes fine we could totally try for the for the for the be hard fork try and go for it and these names that I put on that chart they're just cheesy placeholder names I'm not saying that you should be the names but we need something to discuss it with and you know I probably shouldn't formally name these until we open up the hard fork anyway so these are these are just placeholders okay yeah we can refer to them by the month and year or I mean obviously I like referring to them by cheese whenever we're talking about them for now so okay anybody else have comments okay

thanks for that it's good to put it down everybody's mind and please comment on the magicians forum if you haven't already if you have an opinion on that and July is probably when we should start discussing this again I think unless do you want to sit earlier data unless people object to the six-month timing I don't think we need to revisit this until July when we're ready to start gathering requirements for the next hard work okay sounds good we might revisit this in June just to refresh everyone's memory okay now I think we should add a card for kickoff to the calendar as well hardware for kickoff nine months VIPs at five months so

there's four months to gather requirements and gather finality before we go for the implementations that's probably something I should add to the schedule mm-hmm okay so 4e ip's refer to roadmap link for list that's gonna be the right now it's the etherium wiki under roadmap slash Istanbul and we have 1679 as the hard work tracking meta EIP for now and let me go to that okay that just has some timelines and some rough stuff so let's start with the first one since it's getting close to May I want

to put priority on us figuring out which proposals are going to be going in or not if we can figure out if we can figure out today which one of these we are popular or which ones we can approve for going in we can add it to our list of actual ones we're gonna implement so II I P 1829 pre-compile for elliptic curve linear combinations this also is something that Kasey is talking about later in the call and this actually let's first kind of figure out what what people's feelings are for pre compiles cuz historically

what's happened is we had a different time line for II was um before or something like that and now since things are more up in the air it might be good to put pre compiles in in a way that's a little bit better and not wait for he was um let's see the who has comments on this pre compiled for elliptic curves because I don't think we're cozier yep yeah we talked about this one two weeks ago and I made a comment on the discussion on the forum at Magicians that the specification as it's written it doesn't even have the

cost specification so I mean it's not as far as I see it this is not ready it does it's not specification of what should be implemented so it's kind of hard to to discuss it in the context of should be inserted in the hard work okay and yes so basically yeah that's what martin said is just completely true and the reason for this and obviously the people who are trying to make this happen because I talked to Alex blossom yesterday about it because basically I

suggested him to create a working group on this to actually complete all the work that is necessary for this yeah it should be ready and because basically there's a lot of work and as people might remember back from design team and it's not trivial so it's it it needs all the gas thing and so that's why I wanted to I suggested to Alex to formalize this as a working group and then actually now it's it's gonna be the list of whatever six that I put on the comments that are existing so as a output of that working group we should see the appropriately done the IP whether it will be proposed by May or not or later I don't know but

as I said in Berlin and I don't want to be fixated on this particular date because when if the IP is well specified and properly prepared then the implementation can go much faster and especially if the tester is also generated for it a question from Vitalik because it sounds like his audio isn't working does the pre compiled for generic EC support 512 bit module II or just 256 I think we'll have to put put this question into the working group so I can't answer it right now but let's write it down I'll add it to the cat

herders p.m. issue we'll I'll tag you in in github on the issue Thank You bourse I I think another question just in so Martin I think the question more is are you against this as a concept if the code is good and the EEP is well specified no I'm not against it as a concept not at all but I'm not a cryptographer and for me this is kind of I have a hard time to decide if it's worth it about so it I need it would be good for the people of Champions face to explain to us non cryptographers why

this is so great and why it's so useful and you know motivate like the rebel against it okay great Alexey who's the you said that who's the lead on this yeah so I'm going to probably publish this announcement today because I didn't want to do like multiple groups per day but the leader will be like Sandra blasts off and I will put his details on the whatever we're gathering it and obviously if this is the the working group and then especially if it's actually also at least partially funded then you could expect this kind of accountability of answering that like providing the all the materials and stuff like that so

yeah it's Alexander Vlasov who is a de facto is already working on the implementation has been working on its implementation for quite a while okay anybody else yeah ever comment about not this particular pre-compiled but three compiles in general sure all right a hanger me yeah we can hear you great yeah so historically there's been one case of a pre-compiled that turned out not to be needed so even if the code is good and the EIP is well specified that's not enough justification for adding a precompile so just one example

with the elliptic curve multiplication on BN 128 it was added as a pre compile and because it was assumed that implementing an EVM wouldn't be wouldn't be performant enough and what happened was somebody came along you know exactly Williamson from Aztec and did an implementation in EVM that is actually cheaper and gas cost for I mean for just one one point it's it's almost the same and then for two points or more it's cheaper than the pre-compile yeah now it is true that the elliptic curve

multiplication could be reprice so there's still be an advantage but oblique - so the proposal for a pre-compile for Blake to be is one that I'm skeptical because there's there is an implementation and solidity with a lot of inline assembly and the gas cost is cheap enough to do up to some number of you know of input bytes and I think it's up to the the champion to demonstrate that for one the the use case they want the implementation is eat

and EVM is not sufficient for the sizes that they need that they need to hash and also that they've done a best-effort you know to optimize the EVM implementation because even the Blake to be there's there's opportunities to to optimize that so anyway oh yes oh yeah Thank You Keisha I just wanted to say that it's probably most monstro the Blake to be is out of scope for the for the Working Group I mentioned and the reason being is that they basically choose chose to work on this specific pre-compiled which is generic elliptic curve support and so as otherwise

because basically you can't just leave the IPS and then you know nobody is working that there's no use but if somebody actually committed to work on this specific recompile and I think it's it's warranting the group wherever you cannot really force them to work on other pre-compose because you yes if you find people who are happy to work on others and yets fine but at the moment that I should ask but I think they only mean the elliptic curve pre-compose yeah actually Zuko Wilcox and some of the Z cash people want a champion Blake to be since it's an older EIP and the person who originally proposed it is busy with college so he can't champion it anymore

so what we'll do is I'll talk to Zuko that'll be on my list of to-do items and see obviously this won't go into Blake chubby wouldn't go into Istanbul but just kind of give them an update that there are pre compiles now being considered to go in it could go in to Istanbul if someone works on it yeah I could but okay it's not May yet nevermind yeah it absolutely could yeah so essentially if somebody is prepared to work on this then anything can happen but there so one there's nobody it's working it obviously it's not going to

happen yeah so I'll check with them and let everyone know next meeting let's see what is the next one and I'll try to go through these a little more quickly so we don't use up all the time on this since we have a few other items EW I pee 6:15 static jumps and subroutines that is let me open that one that's from Gregg Brooklyn Powell and Christian it looks like was this discussed last time and what was the what was the verdict I I wasn't able to

read all the notes yeah we I don't think any verdict Brooke did you have something yeah we talked about it a fair bit at the meeting in Berlin there's a few people have concerns about specifically complexity number of opcodes and there's been some suggestions to break a few of them out into separate pieces but for the most part we're still hoping to get the same for the hard work but there's just a bit of a community okay so does this need to be something that is tabled for the next

meeting for decision-making yes okay so moving on from that we'll have that as an action item to discuss next time next one is Prague pal oh boy oh I love this one so yeah we had previously decided to put it in and now I think there was a working group in Berlin about it if anyone wants to go over that proposed so you know whether we put it on its own

test net and yes we've got this point if we were to put it in we need to put star on it depends on the special committee the audit to make sure that it meets its claims and then we'll probably have another discussion actually comes out yes the bad news is like it's the audit might not come out until July right so if it doesn't come out till July and that's kind of the deadline for client implementation I do think clients probably should implement it whether it's part of the

hard work is a separate question and we can do a test net that is just prog pal possibly to replace Rosten because roughs is getting quite old quite large with a lot of strain some live strange branches around some the hard Forks go ahead you know what is the need for a new test that's this is just zero nine three or something else it's updated to zero nine three and none of the clients that have been implemented yet can do both 0 92 and zero nine three and I don't know if we want to encourage that so it would need to be pretty much for starter yeah that's the one implementation right so we also would want to launch that test

that until after the audit comes out and if it's a product I'll only test that that can be launched independent for my main net commitment so if the major clients had it we can want to test that for it whether or not it makes it in Istanbul on an Istanbul timeframe and we can't adjust with the out but the outcome of the auto report is on that regard I would recommend that we don't do it your protests not because we have an ETS front to propel transition and that closer model of what will actually happen on Mena and that's my opinion being in terms of the switchover so that's what I got him did initially they switched over in block 30,000 so I guess

what I'm saying how only test net the purpose is it has Prague now when we start adding attach for a while and make sure that we have a control transition so we can gauge what it looks like when the hash rate transitions over to to Prague house see if it really does stabilize in three to six hours like estimates are okay so the other question I have is why it was on the top of it was on the tip of my tongue oh if we do the Rob's 10 if we didn't Rob Stan and we do Prague POW on their and then we decide

not to put it in like late July early August and just take it out of the hard fork could we then revert back to ET hash or is that kind of one of those hard things to do that's one of the most part needs to do which is why I proposed a separate test net before we start Prague now on because I think you're right here the intent is right that if we do it on Roxton it's tantamount to a main that commitment because we've never done any by test net and then reverted to e accepting contents not always took a second hard work so that's why I think Rob stood until it's truly maenette committed is appropriate croch any other comments yeah change that would be

concerning well we're changing the proof of work we could actually switch so it's because if you if you provide a fast thing to ahead and the first thing you do is verify that head and the proof of work fails then your hope is they're gonna search for the right head instead so we're not gonna wind up in the situation that you're fasting to the wrong the wrong tip and have to go back and forth fast think will still work even if we you know revert back to the former split I don't know if I'm making sense here well I think that would work

on the primary clients when it comes to switching to a proof of work because no real test has really done a solid proof of work shift and they don't have the support for that and they're their genesis files to go to one for period and they go to a second for a period and then go back to the first for period after that so that's my concerns if we do it with that in mind opening extra work on the clients yeah I thought three months if we rolled back you know right you're right we couldn't switch the profile down so it's back to ETS but what we could do switched upon power and as I screw this that's completely where we had seen yes can you just read in yeah I was just about to do that

um well wait sorry I have to get Scarlett to get the whole thing the Telex instinct is that doing another saint-petersburg style backwards hard for contests net is too much extra work it's a bit tricky and because Prague POW is two times harder per hash if he's asking if that's correct I think that's what y'all just talked about right right - you know and then if so a transition is vulnerable to a 33% attack unless there's a checkpoint because the attacker can mine hired DIF historical blocks higher difficulty I'm guessing right that's why I think it should be

you know we discussed a couple of strategies we could apply a multiplier to each new different algorithm say each gets a 2x2 apply the difficulty but you know the difficulties not consistent enough and you know one-time adjustment to reduce the impact of the time frame it's messy to do it at multiplier so it would really need to be a switch at one point to Prague POW and never look back there's really the safest approach so going back and forth really doesn't help okay well we don't have to decide that portion today it's more if we want

Prague how to go forward at all into the next hard fork so is there anyone in this call who is gonna voice opposition to including hard POW in the next hard fork and the reason I asked that is I haven't heard any coordinate opposition that's been loud and I might have missed it but I haven't heard any court have opposition that's been loud and the opposition I've heard from the public and my opinion was that they weren't filled in on this they think it might be a waste of time or they want to wait for the audit to come out as what been hearing mostly so is anybody heavily opposed I think it needs to come back mostly positive for its gonna assemble the auto doesn't come back in

time or comes back in any way negative we need to reassess for Saji oh okay I think that's a good strategy anybody else have comments on that Oh Dana could you meet real quick okay great what was the comment I still don't know how this became contingent on an audit I thought we decided to do this back in January but I'm not gonna go there okay so looks like that one is official for the next hard fork as of right now and that's more it depends on your

perspective it's either a reiteration that it's going forward and it's no longer just a proposal or you can say that it's a new decision whatever you all want to whoever whoever's listening to this you can decide for yourself Hudson just a point of process so we've got the two three three changes merged in so the the wiki is still a decent overview right now but ideally someone will have to champion prompt out and actually PR it into the hard fork meta got it okay you're right that is that is the next step so technically this would now be moved into accepted and be listed in that category

okay later today or next week and all the conditions on that I think it should be brought in on great thanks Dana all right next adding pre-compiled for 8255 one nine this would be helpful for the etherium name service I think Virgil's on the call Vergil if you want to really quick explain how this would be beneficial and anything else you want to add real quick Oh actually you know what he might not have microphone the linear elliptic curve work so there's no yeah no no it's fine the the linear elliptic curve is is a superset of the previous one been basically in short it allows you to to

interact with with with with TLS sister certificates and DNS X zone signing probably the most obvious thing would lie to do so so you know you know how you can get an identity like the HTTP extended validation certificate like when you go to paypal.com it says is it says like PayPal link and the in the address bar we we better verify those kind of identities on ethereal and that'd be that'd be Kennedy question for Jill sure so if 1829 is implemented does this mean that 665 would no longer be

used well need a crack so which which one would you prefer I mean the elliptic curve is a superset so you know I mean like why not have more hmm okay I guess the one thing to put into consideration especially if ENS wants to champion this is that the other one I think as we explained earlier doesn't even have gas estimations and therefore unless it's worked on pretty hard you know in the next month or so may not be eligible for Istanbul and then you'd have to wait till potentially April for the next hard fork so I don't know how soon you all

wanted to send it all or am I am I doing that am ice estimating that right either way as long as as long as it eventually gets in there I guess you don't care which one okay cool sounds good I think we can move on then and we'll just make note that that that as long as either the super set or that one is worked on that's will make E and s happy and they are the use case that we're keeping into consideration for this anybody else have comments on that one okay add a chain ID opcode that one is in last call currently and the review

ends in four days and then we will reappropriation do we should we go over the draft in

progress wait donate the fourth no longer a bra it's whatever oh cool well wait it still says oh I'm out an old blink I think okay cool so that one doesn't have any heavy opposition according to last last week's call then okay we can say that one's going forward then I'd say metallic is your mic fixed by chance because you were the main one who wanted to talk

about the fee market change and if not we can do that next time okay and also we can get Eric Eco gnar and here next time as well to kind of be another voice than this I think he'd be happy to join the call the talaq did you want to talk over text just a little bit about this or does anyone have questions over it I think people might need to look at it more before questions are asked but does anyone have any okay oh there was a presentation in Berlin

about it so that means there's a video associated right Boris yeah time codes are all in the overview page let me go find it great thank you okay the Blake to be F compression function precompile is being worked on by Virgil oh and Vitalik by the way recommends reading the EIP itself and there's threads on Magicians and a3 search where you're he's responding to concerns about it so and then the the link to the link to the meeting time stamp and the or the

time stamp for the discussion and the video has just been posted by boris ryan that's Casey oh yeah and then so they also he mentioned that I wanted to make an amendment to the EAP for 1559 which I which we talked about in Berlin I can just give you a few words yeah absolutely so it's not like decided or anything I guess it's just as something I throw in after listening to italics presentation so you might remember that but Alex the IP is mostly about introducing the kind of two components of the transaction fees one component is called min fee or base fee in different sources and another one is a premium fee and

essentially these are the components of the gas price that first that the transaction Center pays for a transaction and so the one component gets destroyed or burned and the other component gets collected by the miner and even so the first component is B base fee or main mean fee is actually some sort of is regulated by some kind of feedback control in the protocol or in the italics version I think it might be controlled by the miners as well but essentially people yeah their theory goes that it this would lead to more stable and predictable prices the last price is so-so

suggested the amendment which I think would help roll it out quicker which is basically not trying to modify the meaning of the existing fields and the transactions which I think originally IP does and instead introducing a new transaction type which will have a proper format for the for for what italic proposes for these different parameters and so what the way we do it is that we are essentially create two sort of virtual spaces inside the block one of them is for all transactions which we could limit to let's say eight million like a hard hard limit of eight million and then we open up a little bit more space for the new format and so we

allow both formats to coexist and then eventually we will might decide to shut down an old format or do something else so the benefits of that would be that you we could roll out of exchange quicker because we don't need to coordinate with every wallet provider because if they didn't make the change they can still use the old format everything should work but also the second benefit is that if the theory is in what some ways flawed or parameters are incorrect and the the the new format format becomes actually more expensive for the users it doesn't actually ruin the network because there is a safety net of the old format and then but if it does if it is successful then we

essentially are looking at the real data we see when we can start duplicating and retiring the old format so that my there are some some concerns about this my amendments but yeah that was it okay and the talaq had just another comment no need for too explicit arrays just to running gas used counters anybody else have comments on this one okay I'm going to put on my to-do list to invite eco NAR and we'll see that the tally gets his mic fixed by next next

meeting next up we have the page is lost by vitalik he has to catch a plane where's my page on this I think we had one more yes oh this is being actively changed because I think Alex added some stuff so what's the next one we should talk about Oh e IP 1352 if you're on the etherium wiki page for this just refresh the page and

you'll see the updates Alex do you wanna just go through all of the ones you just mentioned they seem pretty non-controversial for the most part oh and you're muted if you're trying to talk yes sure I actually left a comment on the PM issue which has a few more proposals let me just alright so I guess they're gonna start with 30 and 50 to specify

restricted address range for pecan pies and system contracts this has been I think proposed even for Constantinople but it was too late at that point it doesn't really change anything on its own it should have no effect but in the future if some other VIPs are proposed which would deal with recompiles then it can have an effect and so an example for that is a IP 1109 which proposes to while an earlier incarnation of 1109 proposes to reduce the cost of static call when the destination is a

pre-compile and having 1350 to would simplify the specification of such a change because one could just refer to restricted address range any call made to the restricted address range is has like a lower fee does that make any sense any comments I know that the enterprise etherium Alliance are gonna want to support this because they want some address ranges to be set aside so that they can use them for basically

custom sidechain pre compiles oh I think it's a good app and needs people to look at it from workshop it and it doesn't seem that complicated Martin did you have anything ok I saw you and me for a second so I was just checking ok if you want to go to the next one Alex ok the next one is a six eight nine which was added by yuuichi close to two years ago address collision of contract address causes exceptional halt so he this was

overlooked for quite a while and I think so basically it tries to clarify the case when somehow contract creation would end up with a very simple estate on minute conditions but he wanted to clarify this to make testing easier and more simple yeah I don't think anyone looked into this lately but I think it might be might be useful if it's a small enough change we would need a new champion for that since you eat she's no longer participating in

the EIP process anymore as my understanding and we could check on that though at the case I think I think it creates - it may have the case may have changed I think Vitalik proposed this like two years ago and we accepted it something to this effect back when he wanted the original create - so while you which he is referring to his VIP 86 account

abstraction if that would be activated then his change would not be relevant anymore but here P 86 wasn't activated right but I do think we have closed about contract collisions against empty that is implemented in the clients which make contact creations fail I'm 95% certain I can look it up I think we have some test cases that target some parts of it I think what is needed to confirm them it's like it's fully specified and

it's not only specified by test conventions so you have an expect we can refer to so I would say we need to do a bit of research on this but I think it's good it's on the table so we can salvage yeah unfortunately cannot volunteer to champion it I just wanted to highlight the EIP so maybe somebody will be interested in it okay borscht should an EIP with no real champion going to proposals at all I mean the nice thing is that we I think

the one tweak so Alex and I talked about this a bit and and I think there's a little bit more tweaking to be done 2 2 3 3 Alex mentioned this is a wrap up from Berlin that we really should have a deep campion just meaning it doesn't have to be the author but someone that puts up their hand it says I'm gonna work on doing the work that's needed to get this in and so the simple thing is because you have to do a PR to the hard fork oh no one does that hard fork it's not actually proposed so it won't get done so the wiki is an in-between ideally by next cortes call all of those anyone anyone that's serious about

championing something will have PRD into the hard fork meta or they won't and then we can actually refer to the hard fork meta as the canonical list of how things are moving along okay thank you were there any other ones from Alex yeah I have at least one more I wanted to briefly mention it is listed in the draft in progress section on the wiki its 1803 renamed up codes for clarity um so that's again not really core because that's only something relevant to two

languages and tools but basically I want to clarify some of the OP code names too to match their behavior more closely and I think too probably three I would like to highlight I think two and a half years ago there was a proposal to rename the chattery of God to catch up to five six this since has been done I think pretty much everywhere at least in the languages but it hasn't been see reflected code officially so this proposes that the say code official name of the opcode is ketchup

256 and another example is renaming invalid to abort because in practice that's what the opcode has been used to to be doing and I've seen a confusion when people try to refer to it and they say the invalid opcode but then a lot of other web codes are also invalid so this isn't really core again but I think it's useful and the last one is well there are a couple of more in the IP but the last one I would like to highlight is renaming balance to X balanced to be in line with X code size X code copy X code hash because that's the only up code

which doesn't have the text or external prefix yet it accepts a address this first parameter now this is the only change which could also have some relevance to another VIP proposed for Istanbul the one proposed by Martin what's 1884 because 1884 has has an option to introducing you up code for self balance actually I updated it so it's not an option it's earlier on the

88 for was to two variants but then I changed it to being only one variant which has the all right anyone have comments on that no but I did find out in the old EEP 1014 there's a clarification which specifies that the behavioral collisions it's already specified by EEP 684 so I think we could I mean Alex you could look into it and perhaps close the old one from UHG 689 I don't think it's really long

yeah let's discuss this fine I guess on the channel okay sounds good so as regard to this renaming of codes I'm not sure what's the best place to discuss it at logicians forum probably I guess the only the only relevance it has to to core stuff is if you change these names then then II ip's should refer to up codes not only by the mnemonic but also by the actual opcode value to avoid any confusion I would say I would say if

this is not you know if this is not the code change it's alright it's just a matter of basically like how things are named in documentation and tooling and clients whatever and I guess we could just easy if there's no objections we can just easily kind of ratify this and then from now on people would start working on renaming things if it makes sense I guess we don't have to read it I have too much of the discussion about it because it doesn't really touch the code does it no it no it's not anything too hard for but including it and say Istanbul I guess verifies the changes that's the only connection what I'm saying is that

it could be basically ratified even way before Istanbul even like today or tomorrow just like we just do communication from now on we should we renaming these opcodes to this oak code and this is how it should be done everywhere and I guess that's probably gonna be enough no yeah I think Alexis point stands in general a lot of the discussions in Berlin were let's paralyze a bunch of stuff if it's not dependent on each other and this is one of those categories so please work on it and then come back to Cortez when you have an update that says it's ready to go and let's practice non hard fork approvals yeah I think it's really what was that great this is a yellow paper er

I wouldn't necessarily agree with that because we already have any IP for renaming suicide to self-destruct so this would be on the same note okay okay any other comments okay and I think you had one more about unlimited swapping doop instructions right Alex yeah I can briefly just explain that there would introduce at least two new op codes to be able to access the entire stack and

it could also be useful for EMP 6 105 to be used for the cat local set local I guess there's a potential to do that but that the IP right now is still in draft we had some discussions in the solidity team regarding to it and it may need some updates before it's proposed in a final form okay I think that's the last one there and I believe that completes our list for e IPS is there any that I missed

just point order again if you are going to champion one of these eeap's please do a PR in 21679 to add it to the proposed section and we'll practice that over the next couple of weeks and I'll make sure that the wiki ends up being synced with with 1679 but 1679 should be considered the canonical place okay wonderful the next one is item 5 a proposal of a formal process of selection of VIPs for hard Forks this was brought up by puja

and pooja if you want to go through and just quickly explain the reasoning behind this and what this is 1890 the link so the process I wanted to introduce here is like right now what we are doing here we want to propose anything we are adding it to agenda or maybe to the atomization forum or at different places what we are saying that if you need some some EMPs to be proposed please make a pair not be our

case raise an issue a TCH Agata and it will be perfect and then will be listed and then will be discussed as you did today so that was my basic proposal about EAB 1929 and I find that it is a little bit different from the current AIP 2-3 3 which was the question again and again why we cannot be like knotting it together we can merge it together but I feel that there it is better to have it except have it as a separate EIP because the purpose of EAP 233 is to create as since how to create a meta IP for any upcoming hard fork and the target

audience for that may be the third for this like we are gonna doing it for his turn Boyle bonuses championing not championing champion is not the word but he's leading it he is being doing a lot of work with Berlin Meetup and everything he is trying to kind of organize it so he is there and the target audience for this epic 2 3 3 would be the one who would be helping out with meta EIP for Istanbul is 1679 and the other authors or champions who want to introduce their a IPS for this hard fork or for the upcoming hartford

need not go by the EAP two three three they are good by going by EAP 1679 which is already there so what i was trying to convey here is if we have separate small pieces of everything then we can actually edit or append it as per need the flexibility part is there so if we agree that we should have a process defined for doing things like this is very small take for this material blockchain but that this is a small bit can add a lot of value if we define a process how we are going every time then nobody has to be there every time to guide things we will just pick up one

EIP and to that things so that was my basic idea and I would like to hear from like all of you like if it is good to collect the EAP from there and then go ahead with a process like if it is proved right right now we have come up with so many ideas that have been approved yes we are going with that and they would be peered into EAB 1679 and then they would be taking a head well yes I I just want to comment so it's it's very good thought though I would say that we are currently changing things quite a lot with shifting we

actually basically changing a lot in the process of how we do changes and I keep pointing out there to everybody that I am selves are basically not really doing much it's the people who are doing the work so it's actually like their I would start from the from the work then go to the IPS and so because basically putting the IPS there and improving them doesn't actually get things done so I would say that I'm not gonna worry too much about the process specific process for collecting IPS I think we're just we're gonna have enough structure around it for now so and also it is the first kind of hard work where we're trying to do

this meter ap so I guess we probably have enough rules for now for for these kind of things because otherwise I am already almost confused about what we're going to do with the IPS that has to be proposed but still the acceptance this date is it not passed yet we can probably use it for the rest of the EAP switch which people would want to present or want to be included I'm just asking you guys to give it a dry run if this works out from going forward we may not have to add it to every agenda that we want to do it we may not have as much

info if somebody moves to do that we have to just made an issue create an issue mention it over there and in that from that it can be taken over by the person who is actually looking after T hurtful thing like coordinating thing so he knows the window the place from where he has to pick up te IP and and the new IP or the authors also knows where to just write it down just make create an issue and he will be guided from the well it's not like we have like 10,000 people or we're trying to create the IPS to be honest or or to try to work on the changes at the moment we have the opposite problem that we don't really

have a lot of people actually who taking on the work so I would say that it's a good idea but we probably getting ahead of ourselves a little bit here so and I would probably Park it for let's come back to this I mean that's my personal opinion let's come back to this idea a bit later because I think it's a good idea but mmm probably a bit early for for where we are I I can see your point Alexei where I really like this idea I comment in the magician's thread that this should definitely happen but I can see your point where if it happens too early there would be a little too much confusion putting in that process at this point when there's a dozen other

processes being updated at the same time including looking into updating the IP one in its entirety or not its entirety necessarily but with major parts of it potentially so yeah that's that's kind of where I met were I really like the idea but I also hear the concerns of putting it into early any other comics future a be a champion try to amend that and he is looking forward to the end more of the he proposes what they think

and if I mean I was just proposing it to make it a dry land if it goes by it because we have 15 more days over 15 more days to do it and if we goes by that it's fine if not then I'll take it from the next hatful from like July we discussed that they have to stuff that nobody opponent so we can take it for that in July yeah I don't think you really need approval from this body necessarily you can just kind of announce that you're wanting to use this as an added an added thing too new people list of currently ip's active and kind of helped take the burden off people who are updating the wiki potentially with people who are actually proposing the IPS keeping up with the

updates so I think that the people who want to try that can try it and there's no harm in that yeah great so anyone who wants to participate in that just look at poojas link that was posted or talked to her and we'll go from there so that was number five let's have I think Boris and Alexey go over the recent Berlin meetings we've been talking about them throughout agenda items one through five or one through four when it comes to e IPS and stuff but I believe there were more working groups so we can kind of combine six and seven and that'll cover

pretty much the rest of the meeting because number nine Casey's already talked about but you know what just just so that we don't cut off anything else Casey was there something else you wanted to add when it came to the single cross-platform library for pre compiles or as that has that been kind of talked about all-deaf already yeah it was discussed in the comments but I don't see a response from somebody who can speak on behalf of parody so it would be good to hear parody maintainer x' opinion on one question great yeah I don't think they're on the call this

time I think Fred's doing something but they the next time they're on here we can get their opinion or if we want to just ask in the core dev chat that might be better I'll do in tracking issue for this okay thanks Boris okay so now that we've done number nine let's go over the Berlin meetings and the working groups and the potential new working group said anything else Boris and Alexey and others want to talk about so we'll start with Boris and then move on to Alexei for this the technical stuff yeah very brief not gonna take a bunch of time because the links are all in the notes and I'll paste some of them into the the chat

here two days seem to be quite good we just sent out email wrap-up to everyone that attended and that's publicly listed with a with a wrap-up link on the east magicians forum as well proposing to potentially meet again in July so look for that on a magician's and basically feedback is invaluable do we want to do more like these especially if we think about on April twenty twenty hard fork that in fact might be kind of both prepping for Istanbul and thinking about next hard fork other than that I think Alexi or king of working groups as I like to call them should talk more all right King go right ahead

I will go over the my favorite subject right currently which is the working groups and I've listed it for the for the convenience of people who want to look at the the agenda posted the update which I'm kind of going to be reading about so there is a website which is the Ian dot ECM but wiki slash each one we currently have six working groups listed and you might notice that we used to have track like for working groups before I decided well we decided collectively I guess to currently drop their tracking for two working groups we can always resume them but I've listed

the reasons why we want you to stop tracking those two for example the chain pruning it's a it's been proposed back in the DEFCON DEFCON four but they hasn't been a lot of activity and actually we would like to reconnect the swarm team in next month to see if there's we can reform it but with it people from swarm team because there are some people think that there there should be some kind of collaboration there and secondly with the simulation simulation group as we had a presentation from Vanessa at Stanford and you did a great presentation at Berlin and however at least to me it's not currently clear how we're gonna proceed

and we could reinvigorate this group if we after some certain discussions and clarification so the two groups that were there before are still there they stay Trent and II wasn't and then recently we've added another four groups and actually I'm going to add this another one so the finality gadget working group is a I've done a short presentation at Berlin is essentially hooking up the beacon chain which hopefully will launch and this year it's going to hook it up to the theorem clients even 1.0 clients and create essentially finality and from my my preliminary estimates the finality would have a range of basically one hour

around one hour so every every one hour you would basically see a final block hopefully and then there's another working group is we talked about it generalized pre-compile elliptic curve arithmetic sin pairing so essentially there is been a lot of work done and this is just a formalizing this work and I haven't wrote announcement yet but the fact it's already active and pretty much active and actually I got the response about the italics question about the italic was asking whether it's going to support 256 256 bit arithmetics only or or further actually this is very general with you'll support 512 based arithmetic and so on so it will actually they

already have prototypes for BLS signatures as well so yeah so basically the the website which is linked has the leads on it and then we added the IBM evolution group which wasn't announced yet yet but its de facto active by looking at what the work they've already done in researching that I guess we will need to define bit more what exactly they're gonna do in Istanbul but yes so since it's been active so I put it there and the fee market change group has been basically created like yesterday and the lead is a godly and this is to implement to research and implement italics proposal so yeah so that the

idea essentially for working groups as I keep saying is that it's basically the group of people or one person and then later more people coming around to who are interested in implementing something taking it to the end well taking it all the way to hard works basically being responsible for that and accountable for that as well if we pay them basically what my point is that if we pay them if whoever we are like foundation or something so the good thing about finding those thing is because you actually can keep them accountable and then just ask for for results because otherwise if they do it by themselves I mean you can't really do that really keep them accountable so and so the working groups produce the IPS they

generate the tests and they do all the work necessary to make this happen and so the other working group that I want to propose and we just had a good chat with Mitra before this call is essentially Demetri wants everybody all the client implementers to to do some RPC calls and so far the adoption was very very slow and so we ideally need some sort of consensus test working group whose job will be to make sure that these calls these RPC calls are implemented pretty much everywhere I don't know how exactly we're gonna do it but I think we should just get on with it find people who want to do it and just do it

and that's going to be the number the group number seven and that would allow people more easily to generate the consensus test themselves filled them and then hopefully also the more people will use them the more optimized they will become people who have we have a greater attention and we come up with better standards and so on and that was basically my update on working okay does anyone want have any comments or anyone from the working groups have any comments yeah hey this is Zach so I had been doing some work on simulation

stuff I didn't present much of it in Berlin but we've aggregated a lot of times there have been some coordination issues but what I wanted to recommend was combining the simulation efforts with like the testing working group because that would streamline things and there may not be enough work or enough people because right now it seems like I'm the only one that's actually doing anything and I'd only have a lot of time so it might streamline efforts to combine that with like a testing working group yeah that's basically what I would expect to happen so my kind of as we working through this so I realize it's very important for the group to so if

for the group to be successful there has to be a certain level of commitment obviously from the leader and from the members and so we if we wants to make it a proper group we probably have to find a leader who can dedicate at least half of their time to this or ideally the full time so but let's talk about this I think I'm glad that you know we come back to this and then which let's discuss this to do more there but also I think that some of the simulation efforts will overlap with testing as well and me and Demetri had a meeting a couple days ago talking about testing

and forming a working group so I'll let him fill in on that as well probably with testing updates okay I think one of the things that that came up but it actually came up with the see market change that the vitalik proposed when he gave his presentation was that and Alexi kind of talked about it a little bit Alexi is have some funding for the etherium foundation to do work on these items and that's the kind of elephant in the room is that if we expect these working groups to essentially do research on each one X and work on implementation or in areas

like infrastructure like testing then there needs to be money just to say that we ran out loud rather than resources or something like that to support the serious commitment of time to these things a lot of us have been volunteering time and are happy to do that as part of a large open-source project but there's a certain level where of commitment where that has to be compensated so that's obviously an ongoing discussion if yeah if you'd like I can kind of be a liaison for discussing with the etherium foundation potential funding requests and things like that and kind of be a yeah like a

conduit for communication there so yeah just boars talk to me about that later as that get a comment yeah I was just gonna say also like keep in mind like at least speaking on my behalf as you know I work at white block full-time and we're a for-profit company so I wouldn't think that it would be entirely necessary to seek a large amount of funding all I would really need is essentially a junior developer that can kind of take direction from me and that would be a cost that I would also be willing to cover on behalf of white block because I could put that I could

write that off as R&D so we can be pretty efficient and optimal with some of these efforts and you know at least on my back cool that sounds great any other comments that's a really good overview thanks everyone next up on the agenda would be we already talked about the prog how odd it update audits funded or the audit is funded so testing updates would be next I don't think Dimitri's on the call did anyone else have testing updates Zack went over the fact that he's having

meetings with Dmitri but did anyone else have like specific updates Martin do you need help with high test or are you gonna coordinate with Dmitri and Zack yeah yeah I mean working with Frank sounds Lord's to implement some stink tests and some peer-to-peer peer-to-peer tests for clients I've been in touch with Zack I'm not pretty sure what what collaboration there will be they're kind of building this generalized version of hive yeah

ok I think that's it for testing updates then I was just gonna ask is there a good way to time box items is there like a method or process that I haven't read about before for time boxing that anyone has a recommendation for or even just an original idea and that would be during these calls by the way whoever is leading them make sure that we have enough time we didn't really good about time today but in the future if there's more items we might need to time box all right more of a hard-ass I got it Boris she's saying that you know I just

got my opinion like late last night Pacific time and so if we can just get some of the things in earlier or even have discussion ahead of time you know who knows how we can do that but that could maybe just help have a cadence where the discussion happens and we just have questions on the call rather than having to go read it and then discuss on the call so that's just improving our process ok good feedback I think that that's those are both good ideas does anyone have any more all right cool any client updates I didn't see any specific ones and the comments that anyone wanted

to talk about as far as I can tell oh well I know I just wanted to I realized I forgot that there I did a little comment until we get we say basically I'm just keep using it for the stateís client prototype and so unless somebody if somebody hasn't seen my latest publication about this shade of faithfulness essentially you know you could look it up and I'm trying I'm gonna try to write more accessible version of it but essentially I'm coming to the idea of making the flexible trade-off between the statefulness and statelessness so essentially to have the

clients can dynamically shape-shift between being stateless and stateful which is quite interesting so um yeah that's it and then Matthew I believe the last name is Halpern said I can't attend the meeting or I'll be listening to the meeting but can't speak and it looks like ETH version 64 was presented at the one point X meeting without much commentary after touching base and getting updates on the sync progress I think it's best to wait until these mature before locking anything in he's gonna aim to get a IP drafts out before the next Court have call and continue to track the sync progress over that and that is a net that's the networking

protocol right yes correct so I can comment on this because I have been talking to Matt about it as well and so yes there is a certain overlap between the proposal for the East 6-4 which would be the next version of the theorem what we call it I mean it's I don't know the correct term but this network kind of one of the network layers protocols and you know and the andre was also presenting but berlin were what workshop the first sort of spec for the the red Queensland protocol which is for example introduces read request IDs and every in every word operative and so the same

thing happens in two if 64 which is actually kind of a Moebius thing obvious improvement because it has lots of lots of performance implications so yeah but I think it makes sense to forget it sorted that get specification and to make sure that we have correct overlap between those things okay anybody else have comments all right I think we finished up three minutes early unless there's ewaz more research updates but I saw Danny left and actually the other day I said

only if they were posted in the comments below in the agenda so anybody else have any other things to say great Alex I just wanted to ask for clarification on

the CD hard meta Boris mentioned earlier on the call that the the wiki shouldn't be if I gotta correct it we shouldn't be the main source of information rather they had fort matter should be the canonical source for these discussions which yeah peas are proposed and as of today on the matter itself it's a free and me listed SD author's so that means we are the ones you can click approve for the EIP autumn urger so i just wanted to clarify is is that what we agreed that anything can be proposed and i just gonna click yes if it's

pointing to a proper AIP yes that sounds good to me lots of things will die at the proposed stage I think one more tweak that we should make is that this concept of a champion should be listed there as well so the only one that's in there right now is Bryant with 1344 the the chain ID op code um and maybe if no one objects Alex I'm happy to essentially help manage that for you since a freeze no longer involved if we want to make that change yeah what do you mean by the champion we

have like have it listed on the on the Harford meta or have it listed in the in the EIP itself the hard for Mara yeah I think I think that might be a good idea so at least people know who to contact for questions yeah we can also take this discussion offline one last thing I wanted to to mention in regards to that I've created a topic on the magician's forum for the Hartford meta and added it as the discussions to the URL in it I haven't found any other topic which is specifically for this but it may be a

good place to discuss which he at these people proposed to be included and it might be a good good forum to have discussions only for the proposed VIPs and maybe the transitions when transitions are decided on okay it sounds good all right thanks everybody we will see you in two weeks bye happy Friday [Music]

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

[Applause] [Music] [Music]

[Music] [Music]