All Core Devs - Execution #63

Transcript

[Music] [Music] [Music]

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

[Music] hello everybody and welcome to the core

dev call number 63 I'm gonna just pull up the agenda here it's gonna be the agenda number 102 on the URL and the first thing we're going to do is review previous decisions made in action items if you click on that link so there weren't any decisions made that we need to go over specifically just a lot of discussion over some of the e ip's some of the actions we did put down that we want to check on reach out to the authors of e IP 615 to discuss the questions around the EIP especially the option of splitting the e IP and the smaller more digestible components so that is who's the author of e IP 615 is

that I had that's great Oh Brooklyn and them okay did Greg respond on get her at all there was a lot of chatter on there so I didn't catch everything don't know if anyone else caught it either if not we can reach out to him specifically for next time I don't think there was much more discretion specifically around that past the AMA a date where they talked a lot about all of those let's go ahead and just table the CIP then it sounds like there hasn't been attempts to really push it forward so far if anyone wants to check with

Greg that'd be okay but I haven't seen real attempts from the authors to push it for yes like I either from conversations with during the guests and of that yeah that's not really bringing any improvement in the short-term and yes so they want you to integrate it immediately well okay yeah that sounds good in that case yeah we'll table that one

and see in the future if they want to split that into smaller pieces and over time maybe implement other other other things on that that are more simple so 62 point to have an in-depth discussion on e IP 663 let me see which one is that unlimited swap and doop instructions and that was from axe ik and I believe AK 6 is the on this call yeah you are did we have more conversation on getter about that axe ik oh and you're muted says he doesn't have a mic ah okay you

can just text the chat there wasn't any progress made on it okay we can work on that more since it sounds like there's gonna be a few e IPS that are gonna need that to happen can you lead the discussion on the all core devs chat just to see what the temperature is about this e IP so we can do a final decision on it as soon as possible yeah he's typing in the text chat okay he confirmed that he can he can put that on his to-do list all right the next item perform a full benchmark

of e IP 1108 so EIP 1108 I'm just like googling these as it comes up that's gonna be reducing the alt b and 128 precompiled gas cost and I know Antonio and Zack Williamson were working on that specifically and I remember us before talking about them doing a benchmark or that either of them on the call today I can actually speak for Antonio but Zacks primarily been driving this latest version of the heat I guess it would be helpful for me just to know what you guys wanted to see beyond the

benchmarking that's already been done um I think I wasn't on the last call where this she was expanded can anyone fill us in on what that specific action item was supposed to entail okay well I don't want to take up too much time you guys don't have anything but I will all talk Dantonio and circle up with Zack as well and and hopefully have some progress for you guys I'm getting up sometime next week thanks and have them just reach out on the all core devs Gator channel so we

can keep up with that and take notes from there that's gonna be the most efficient way to do it thank you yep discussion required with Geordie ballina and Al ataxic around the options between a IP 1109 and e IP 2046 so that's between 1109 what was in 1109 actually Geordie you're here aren't you yeah yeah okay could you just quickly go over the what those two e IPS are and if a conversation has been had and decisions made over which one would be

the better option to go with for the upcoming hard fork if we decide to do that at all Hoffmann check at least new one I can't explain for I can talk about mine mine is a very simple mainly is the the cost of the smart phone of coming up recompile smart contracts is very expensive right now and there is to reduce first it was an EAP that tried to reduce the cost of the call or the static call but the developers of the of the other of the clients they suggested that they didn't wanted to twitch to touch the call code and it was better to implement a new a new approach for calling smart contracts and that was

that's the work that was done in 1109 and I have no more comments about that okay and then exit if you wanted to type anything no discussion with Jordi in the past month regarding this okay so we'll keep that as an action item if Alex and Jordi could get together on and the core devs chat or an 1/8 magicians form ax ik also said that 2046 is the original version Jordi mentioned so it looks like 11 o so 1109 s the newer one is that right even though it's an older

IP number is 1109 like the newer one or is that the older one like as far as which one was created first 11:09 was originally the one that knew a new opcode I believe I don't know if it's been updated to just supersede the later number understands what's dong because you know that the core depths are in a call in accordance they decided that it was not it was not good to teach them the code of the cult code so that's why it was changed fully changer to

these to introduce these new oppos understood ok x''k what's your position on that is that something that you'd be willing to agree on in order to go with what was it 11:09 rather than 2046 am I getting that right everybody I might be getting that backwards you're getting it right okay and just to note that this is one of those the three or four either/or decisions that need to be made about the hard fork ok so we need to get more people on that and as we're going through these if you if everyone goes

back to the agenda there is a link under to point a about the dependency map that James Hancock made as we're discussing some of these action items you can actually match the e IPS to this dependency tree which is really helpful because it shows what needs to be decided somewhat and it's in a distinct order in order to get certain AI peas chosen for the upcoming hard fork because some of them are basically incompatible with the others as then there's like 4 e IPS and we have to choose one of them back to the action

items SP what the new and I can comment on that that the problem of the of his new proposal of 2046 is that static calls you need to load the code you don't need to load the state but you need to load the code so this so you can reduce the cost but you can not be loose as much as it would be recommend that for recompile the smart contracts think that there are pre-compile smart contracts like sha 256 anywhere some of them that are lower that you want to have a very low gas cost we are talking about three four five six gas costs some of them may be

so this one it's a static car will have to you cannot go that blow of the in the gas cost by just reducing the spectacles because you have to load the code of a static okay so you're you're also so is what you're suggesting to not lower the cost what I'm saying is that I will not touch the static call as it is right now and I will create a new of code for that just because of that because the static it works for any smart contract and you need to load the code of any smart

contract so you can have an attack that consists of having a smart contract that caused many different many different contracts even if they are static and the client will have to load all of them so you cannot reduce the gas very much with a static column so that's why I think it's better or a better option to to just go with a new of code that's a specific for free compile to smart contracts and then let that cook okay it's just to you know or some normal minimum gas cost like another instruction okay thank you for that this is something that you'll need to discuss

on a specific form and coordination with ak6 so that you can lay out your viewpoints and people can discuss it so that we can have more coordination for the next call and we'll need to y'all can talk on the all core devs Channel that'd be perfect to say do you want to do this on an ethics forum or set like a specific time in the all core devs chat some of the e IPS have been doing that and that seems to be working pretty well so if you could coordinate that with AK SiC that'd be perfect ok thank you so much and I can help I can help coordinate that too Jordan Oh awesome James yeah ok so the next one is 62.5 James Hancock to update

the wiki and meta EIP 1962 with decisions around the EIP s I think the meta was updated right the meta EIP there the meta e IP there isn't really much to update on it anything that's really moved forward past that point that is my understanding there are some things we could say are more soft-nosed and softer yeses there is unless someone else has other feedback that's kind of what I'm feeling like as well but it's good that we're making some progress with the soft-nosed and soft yeses we'll

keep that action item on the board just because that needs to kept be kept updated and we can check in on it for the next meeting as well alright so the next one is that a IP 1283 requires a new VIP number and section discussing the difference between the original EW 1283 which was removed from the Constantinople and this new e IP so there's that the one from way I'm let me see 1283 okay and does there need to be any changes to

the to the authors on there if I I think if I recall that Oh cuz it has Alex and Y do you want to be added to that authors list way okay yes okay you mean either way or okay sounds good so the decision made there is that we're going to have a IP 1706 take over for a IP 1283 and the IP 1283 will be withdrawn way if you could go into e IP 1283 and do an update to change the

status to withdrawn and just leave a note in there or a comment or something that witchy IP supersedes it that'd be excellent it's right and I I'm not sure like how that part because technically I see so then I guess if you're applying both a ip's should everything just be

combined until one is that kind of the way we should go or do you have a recommendation just about documentation but I don't if we just say we Poe CIP again 1283 and 7006 that will work because most clans like like if like implementation for 1283 now didn't any change we just need to add an additional condition

cause for 7006 so it's kind of if we said config rameters in the planned implementation is actually quite an independent I just we need the we need 1283 to actually implement and 7006 to actually implement complete net gas meter this is just from Alex that he he recommends that combining both of them into a brand new VIP so then 1283 and 1706 would be superseded by this one is recommended

I'm reading the chat for him because popping up all right thank you yeah I I agree with Alex that it would be a little less confusing if we combined it all into one with that so but it sounds like would that provide complications in the code for the fact that 1203 was or 1283 was already implemented i because this is just about documentation but like my i have a tendency that probably cave 1283 and 7006 would be better because i mean at

least for party it will create some confusion about which yet he number we actually used for the config parameter because right now we can just implement 1283 and either dollar config parameter for 7006 but if we add a new new yappi numbers and you have roughly the same functionality of 1283 and 7006 and is uh I mean a little bit a little bit computation for transient implementations in 1706 my 1283 you just turn on 1706 and it gets 1283 with it in

the code or is your code not set up to handle that elegantly oh sorry I didn't quite hear the question can you set it up so when you turn on 1706 it also turns on 1283 I'm whether or not there's a flag for it or is your code not able to handle that elegantly and it'll look like I can they got a key yes well if you handle an elegantly like we we can have like like if we can turn on 1283 and around seventy or six separately so an issue is like the so it's it's the the because the changes in the code are in completely different

places just we need those to ya be combined together to implement a transmitter this is from Alex that probably getting input from the testing team would be the deciding factor and that might be a good action item to fall back up on next week okay great so between now and then James can you coordinate with way and potentially Alex and Neil AB from taboo key to talk to the testing team or even just talk to the testing team yourself and figure out what's this oh I'll reach out to Dmitriy and see what he says okay that sounds great let's go to the next one run through these where was I

engage wrote in Sanford and bride Eisenbach to discuss 44 which EIP 1344 1959 or 1965 should be implemented are either of them here today Ronin or Bryant okay I think Bryant said he was gonna try to join and then I haven't heard since so we'll keep that action item on the board and try to reach out to them again action items 62.8 the IP 1352 needs more work done to answer questions posed by the all core devs VIP 1352 is specify restricted address range for pre

compiles and system contracts and that's from AXA qug n he said there's no progress from that so we'll keep we'll keep chugging along on that make a lot of their range and with the registry so their support outside of court theory for precompiled range and general registry for these and if it existed they would use it for their permission contractor specified in the EEA specifications does he want to remember what the opposition to this was I think it was that there was no

technical impact to code oh so this would just be instead of a quarry IP this would just be a meta possibly I mean there are some things that it could have worked with the code if the precompiled calls knew that there's only a range it would make verification quicker those are some hard fork features we can add on a later date and if we just met a reserved range for now and meta registry for these we can go for the technical implication implementations that would help verification since we're not doing any sort of a verification it looks like you know Stamboul it does right to the level of just a matter right now is there anyone who's opposed to this because it

seems like a pretty easy go ahead so it's supposed to be easy but there are some edge cases that have to be clarified for example if like what happened for example do we allow too like sexy dresses if I any like the others will be much the restricted subset and I think it was Martin ended up requested well ask some questions have been requested to clarify that well I'm guessing here the toxic toxic meant by no progress okay let's try to clarify those before the next call and that'll

be something specific we can target especially with conversations with Martin because he's been active and the all core Deb's good arch at the next action item is that is it e IP 2045 needs further discussion and that a I let me type that one in particle gas cost for EVM opcodes and that one is by Casey de trio and Alex pairs Ozzie Casey can you comment on this has there been more discussion on this or any kind of resolutions or acts akin the chat either one here can

you hear me yep I could hear you yeah I guess in brief the next step for this one is to do further benchmarks on EVM one and and propose what the new guys costs would be four for each each opcode another thing we'd like to see is other client teams such as death and parody taking a look at the optimizations used in UVM one and trying to port those optimizations to get to defin parity is this something that we should table for

after Istanbul in your opinion based on the need for more of these benchmarks and collaboration with other client teams that may be busy um well no that's why we submitted by May 17th because we'd like to see it in this turn goal okay well we can keep looking at it then other client teams is this something that can get done before the next all court devs call is there an appetite for wanting to do this unless there's a gas cost adjustment that would bring it down inside of this temple I don't see the real impact and I'm not seeing any gas but yeah that's what it

is it is if we need to do the benchmarks to then you know do more thorough benchmarks to see what the new gas costs would be but we know just roughly that I'd probably be you know for computational opcodes the gas costs would be at least ten times cheaper so when we put the schedule together May 17th was supposed to be the EIP is already and I would have expected as benchmarking done before calling an EIP ready to be honest yes yeah that's kind of where I there's a lot of placeholder VIPs so for example the Blake to be was not ready ahead May 17th and they're

just now refining it so you don't all use the same standard yeah and I think then also the per the person who was going back and forth most on this was Martin's and since he's not here today I it's the in in the in the all the court and the AMA that though that was the primary discussion was between those parties so without him here I would say it'd be good to keep the discussion going okay we keep the discussion going as far as what Danna just brought up James because you've looked at all these IPS over and over again thankfully to make charts and other pieces of data what's your opinion on the readiness of

some of these II IPs and if we should stick to a more stricter set of guidelines and be a little more assertive about cutting off some of these that weren't completely ready is it was it the case that the vast majority of them weren't completely ready in the same way this and other AI peas like Blake to be were I would say most of them are and the discussion at the time we the week before that the before the deadline and the call before that deadline I think you may have missed that that call was I might have I was is that this is just to get like the placeholder idea is okay whether whether or not something like there's still a lot of work that needs

to be done to get everything in and the work is now on as on the EIP editors to get that work done and service VIP authors and on there and client developers you know implementing them and so forth there was a call that Tim chaired and a lot of the pressure to just get the ideas out was coming from Boris who isn't on this column right now well and actually from Alexei and Alexei yeah he was versus trying to get Alexa to get the stay read stuff from Alexei Alexei didn't feel he was quite ready so then he had the the place hold the idea is just getting the placeholders in and and the currently the to get something

merged to draft it's basically does it does it format and does it have some kind of spec like that that alone is really I just feels like going a little bit backwards on what was already decided to say that now yeah I agree it's going backwards so it's something we can't do we will have to figure out this conundrum of having so many a IP so we to avoid this problem the next time and just have stricter standards and that's the bottom line I think Danna what are your thoughts yeah we can go I'm because it was expressed and very well socialized but I think on this this illustrates the problem in the

next round when we go for the hard work that there is a VIP ready date which is what I proposed a VIP proposed date two months ahead where we have this game where people put their good ideas out but then they have two months to finalize it yep yep have to do that better next and the current was just to be draft with the requirement and then even that was a little bit more wishy-washy at the time so that which around the specification on what that means would be important if we're gonna get anyone change okay questions about the fractional it does seem like a lot of work for a value that I'm not immediately seeing suppose benchmarks would be very little maybe

we have benchmarks in the US and slash benchmarking repo roughly of the speed ups compared to death and parity but even if you just compare the benchmarks by time the there's reductions to be made even against the death and parity speeds so then there'd be additional reductions and yes we're working as fast as we can to put out more thorough benchmark reports and hypothetical yes analyses and so forth so could you also put in there what's the harm if we don't go to fraction locally Florida at one

could be inflate the others does it have that much of an impact on the style and level of smart contracts the more elaboration on that I think would help to yeah I mean that's a briefly mentioned in the EIP and it's a complicated issue because it relates to you know what would the consequences be of increasing the price of S store and S load you know in to enable boosting the gas limit without increasing the rate of state blow so this is an alternative to that ASIC and Chad said the main benefits the potential is the potential to implement a certain pre-compiled proposals is Native EVM code so I would

assume that we do EVM instead of the pre-compile yeah I mean I would say there's a lot of I mean there's a lot of benefit it's not just it I mean the gas reductions are so significant yeah it makes implementing certain pre-compile that as implementing them in EVM and still be cheap enough and I mean in general all computation will be a lot cheaper so could we just shift everything to left by template for theirs I'm not sure I get the joke ten times ten to a thousand times more and then we get the

fractional from there that was Martin's feedback and the AMA was making one be something and increasing the rest and then you would need to change the gas limit too well Martin took something a little bit different in that all right you know I'm yeah anyway so just shifting the gas cost that would I think that would have an impact on all wallet and user transactions and you know and the way gas prices are done only on the wallet side which we you don't want to you know we don't want to affect that can you add that to the rationale it is in the rationale Thanks I mean I just

want to comment that like if we do it in the house and have this um sorry I was trying to follow but it's stability so it could have some sort of like the reentry back like we have for casino bow so we may need some conversion or other sort of precautions to make sure we don't have any vulnerabilities but instead if we reach the gas cause of either of course instead of reducing the cost and we won't have well so the

reentry bug was caused by reducing the gas cost of a store which we're not suggesting we're not going to suggest it's why it's only for computational opcodes now there it's true that there might be similar cases what have similar you know negative effects that you know side effects that were not that we'd like to avoid detailing or investigating and analyzing and figuring out what all the invariants are that we'd like to preserve and which and if gas cost reductions would of you know state computational opcodes not a store would break any of these environment

invariance yeah that's like this further study analysis needs to be done on that but we also know that raising the cost of s store would also have would also break many invariants so but so you know that there's this variance that raising the cost would break but there's not any that we know we're lowering the cost of computational lab codes there's not any known side-effects from that but there are no one side effects for raising the cost of the store job cards hey we need it we we need to take this offline

because we have a lot more to get through today sorry for interrupting but if we could take this to the all core devs chat and is there an F magicians forum Casey okay post that in the hall core devs chat in a way if you can jump in there maybe Dan add to and sort through this I'll we can have James also talk to Martin and others about getting in there so we can figure this one out before the next meeting hopefully great happy - thank you yeah online offline there was someone in the comments of the agenda that said they had to leave early and I want to respect

their time so who was it that said they had to leave early and want to talk about something before I go through the rest of the action items I think that was Alex blossom and you had to leave earlier than you expected much time here to represent some of the topics of 1962 instead got it any are you gonna be here the whole meeting not really but I'll be here enough to represent them okay um alright if you have some that you want to go over specifically let's just do it now that'll be easiest sure awesome so I think I don't know what has been been

discussed before in the chat but should I introduce the IP or with that yeah go ahead and introduce it down just before we do that though is this something that's gonna be an Istanbul or is this a new VIP so I think the hope is for you to be there okay and it hasn't already been proposed are you aware yeah so 1962 is there is okay no problem oh this is the easy arithmetic comparing with runtime definitions right yeah if you

can do you can spend a minute going over that and then we'll see it all right where we can best move the discussion cool so precompiled that allows for efficient operations arbitrary elliptic curves that I represented in short virus transform it supports four kinds of operations it supports elliptic curve additions with multi exponentiation is kind of multiplications like single ones and paring operations of some of the elliptic curve families that are known to be good for parents so that's the

background what's the current status is that Alex's implemented all the proposed spec in rust and there is a report that implements all that together with this binary interface that this pre-compile will accept and currently I'm working on creating an alternative sage implementation or let's say collection of sage script to test for correctness so we're generating a bunch of arbitrary elliptic curves and trying to do some

gas metering and say that things work well and try to come up with gas costs I [Music] think that the main messages that he wanted to pass he wrote it as comments in the in the agenda is that there is a desire to expand the interface for all the operations additions multiplications multi pronunciations for not only for the G one group but also for G 2 and currently there is only one

implementation that alex is working on and there is no one yet that expressed a will to make go or C++ in implementation and think besides that it's pretty complete from his side I think and I'm trying to make verify excellent I see this is very complete this even has benchmarks and some test cases already in the code base yeah so that's excellent there is a discussion thread on eighth magicians and Coby if you can

go to the all core devs chat and just some time next week maybe early next week just prod the chat to go to that forum and discuss it a little bit that's gonna be the next step forward for this the IP sure okay thank you and yeah if you could also let Alex know since he wasn't able to make it and this will be another one on the list for me or James to catch up with cool was that the only one from you for now or was there another one you need to represent Alex for and that's the only one okay thank you very much 18:29 it's one or the other how does that

debate dan so I'll tell you what I know about it but maybe there's someone else has more opinions so 1962 can supersede 1829 it includes whatever is there but it include it's a big superset of it it includes much more so we'd get all of the one that it's superseding if we did yours so yeah exactly the 1962 is the one that is superseding yeah got it okay Oh 1829 s the pre-compile for elliptic

curve linear combinations and if I recall I think Remco was okay with that but I could be completely wrong so we need to reach out to rim Co as well that sounds right to me - okay we can look in previous notes as well okay thanks everybody action items sixty one point one the ganache spectra second yeah I thought I was gonna sneeze sorry what was it James we it's over sixty two point oh we did the IP 2024 discussion required between James Casey and Zachary what's this update on that guy so we went through

the IP 20 24 a little bit we you know where we left off last time was that we wanted to go back and benchmark you know proposals here and so the key team has taken on technical champion for this and has put together a great proposal in VIP poll 21:29 so this goes back to the original AIP 150 to take on it where we want to implement the s compression function instead of the whole Blake to be function this gives us a lot you know smaller changes to the code base as well as letting developers use this news all of Blake to his features without needing the core theory embeds to support all of

those so keep has put together an implementation and benchmarks for this which you can find in their repo they already have it you know building in guests I believe and so the benchmarks are extremely promising s compression is like taking nanoseconds for often and basically the F compression approach is superseding the straight Blake to pre-compile approach right now Matros posted the links in chat for everyone probably chains Hancock James I've been

working with them on that yeah it's it's definitely look forward for both the etherium side and the Z cash side are looking forward to be able to do interoperability the kind of the last question for the for the group is should 2's also be included as supported and if that is then we would need to add a separate recompile for that as well so we might think on this I've spoken to the handshake team and he saya team as well and they're very excited about this no production chain uses 2s honestly it's almost never used anywhere that I'm

aware of and I don't think that it's a I don't think that we get enough benefits from implementing to justify a pre-compile there's not using Zika it's only used within the joint splits and shielded proofs it's not used in any other part of the consensus and because we essentially can't verify those in aetherium anyway reliably I don't think we actually get real benefits from it slightly but you would be able to verify them with the the new elliptic curve frequent piles and it will hopefully also make it into a stream dual verifying them within etherium doesn't provide any benefits to a theorem

contracts because we're limited to STV security the same way verifying Bitcoin transaction signatures doesn't add any security to BTC realize ok that makes sense I think there might be one use case for Flake to us though which is reusing the same Z cache circuit the new ZK circuit in you know in in etherion contracts and verify snarks with the new elliptic curve recompiles but so Matt helpfully posted a link to Zuko and the Z cache teams take on this and they're still saying that it's a nice to have in

their opinion they don't think it's necessary I also think it would be a very nice to have and I'm not saying we should never do this but for the purpose of like this discussion we have a great implication of the blake to be f compression function we have good benchmarks and i don't want to let 2's block that yeah fair enough yeah I like to see the good progress but seeing the progress on that does the IP we're we're happy to extend to 2s and I think I speak for the keep team as well when I say that we can get this done and in but I don't want to make it a blocker for the EIP process yeah I'm not seeing a lot of people speak up for 2s

um extend the only one we can think of is that Z cash one that Casey talked about so I would think that because the Blake to be people already like the one that James Prestwich is recommending that we should go forward with that is there other opinions in the room about whether or not this should go into Istanbul or if it would be hard to implement things like that cuz the IP seems fairly complete from what I can tell and we could and it would help a lot of different pre-compile that are wanting to be in there from different projects I would move for if it's a nice to have that it should go into the next

fork when it becomes a more than nice to have go ahead James Prestwich sorry I was just gonna answer Alex this question in chat real quick 21:29 is the most current version and now and I'll put those all together into one and get them into the get them into the right format wonderful I see this as more than a nice-to-have though I see this as something that would satisfy the longtime want of the Blake to be being in there and then the recent want of a handshake and SIA and stuff like that wanting to be kind of adaptable to

aetherium I was meaning the the Blake to s part of it yeah the Blake to s is the nice to have the Blake to be is the must-have for all of those things to happen and perfect one of the major reasons that we moved back to the s compression function approach is that it can shake em saya both use customized variations of Blake to be and so using the eff compression functioning it says maximum coverage on any possible Blake to vary play to be variation sounds great to me who else likes this or who else doesn't like this I like it so we put to be on a stand goal and we

put two s on the April 2024 to be clear you're meaning James Prestwich is VIP and not the to bei P that's alone right sure yeah okay yeah so we put the one that's all encompassing and then sv or two s can be in the future is my mic sounding better now by the way I tried to fix yeah great you know those like ones that like stick out a little bit and you have to like adjust it it was like literally up on my head and I didn't notice it so is like getting mic sounds from my hair probably so okay let's go forward with this then it

doesn't sound like there's anybody opposed to it and it would benefit the ecosystem a lot and I'm seeing a lot of people say yes this might be the first di P that's actually a full on as totally approved forest and bowl so that's great and Alex said something about there's a longer discussion regarding the benchmarks and the feasibility of like two and an EVM and I can't once I get that things can once I get the VIPs consolidated into one place and then looking at the benchmarks I can send those and talk with you about that Alex okay great um next we have item 61 point 1 the ganache spec compliance

issue it go aetherium there's anyone from ganache here who can speak on that okay will could table that one and keep it on cat herders to look at updating the IP one that's PR 1991 is there any update on that from the cat herders so a lot of people have some really good feedback I posted an update to like based on on people's feedback last week so if anyone who's interested in this once to have it it there's a newer version up now excellent we'll keep that on the actions required list so we can keep looking

back on that and we have made it through all of the many actions required and we can go back to the main agenda now where we'll run through the rest of this okay so going back to the e IPS and Section two of the agenda we have 1962 what did I do here 16 I meant eeew 1679 but did I also mean 1962 someone can speak up if they want to speak on 1962 I don't think I meant to put that there 1679 the hard fork Mehta we already discussed that under there's the dependency map take a look at that

everyone and if there is something you can contribute to solving part of that dependency map for instance if you see any IP on there that's yours but you want to withdraw that would help us clear up that map a whole lot and we're also gonna need to work on that little by little to see which EIP s are actually gonna go in that's too much to do all today though the etherium network upgrade windows a I P 1872 which is a meta EIP was danos Danna has there been further discussion on Earth magicians about that I think we've kind of decided that the next one would be April kind of is that what this reflects make sure there's buy-in that we ship to a

particular third Wednesday of a month to provide flexibility for people planning vacations I mean the allusion to this is patched Tuesday a lot of IT folk will not go there on vacation on a week continuing the second Tuesday of the month of case Microsoft releases real big sneaker that they have to fix so by you know supporting this schedule that we release on the third Wednesday of the month the node operators know that if something is gonna be rolled back within 24 hours they need to be on that on the job during that week in case things go wrong ideally it's gonna be a boring week but it's like landing an airplane it's either boring or terrifying got it what do people think of that I love the idea I also like the idea but

just to echo Martin from awkward ev's getter this week I think he had some concerns about we perhaps want this to be an even smaller window so yeah I sort of highlighting that there was still some discussion around do we want like single 'part forks and that type of stuff yeah so there is flexibility the road map the big deposit ones would be every three months but there was a priority upgrade that could be any third Wednesday and of course the critical is gonna be whenever we need it but you know the question I guess is do we go away from these larger hard for it clusters or did we just ship every third

Wednesday with whatever we have I see and I know Martin was really vociferous about that and I would add Alexi has been on has also had that echoed that the shorter is more okay so this needs more discussion yeah so we'll push to discuss this more on the 8th edition forum and luckily this isn't something that has to be decided quite immediately cuz we're kind of in the middle of this stand below let's table it for a month ok sounds good to me next up we have the IP 1109 precompile

precompiled call opcode or move call cost for precompiled contracts did we already go through that it sounds familiar it's just another one that's a pre-compiled call up yeah that's we went oh that's Jordy that okay okay I was really scared that was a third or fourth one okay so the IP 695 is one if you click on it it goes to a comment below that Felix wanted to discuss about a minor RPC spec change regarding RNs signature values everyone is J Carver seem to like that idea is Felix on the call no he's he's out until next week you'll be okay we'll

just talk about this in two weeks it doesn't seem pressing to me in fact it's actually just a network e IP so it's not even an Istanbul but it is one we wanted to discuss and finalize okay next up we have working group updates I don't think Alexie's here own the links broken great about how some of these gas costs might open up other bugs like we had with 1283 that we just don't know about and one things mentioned on the kinder chat is that we had account versioning for all gas prices change we and have each frame applied to the

version that the account is within then that would isolate most of those issues so I want to make sure that we get some consideration of account versioning for Istanbul James you had something uh in the just from the AMA there was a lot of quiet days and a lot of loud days and the loudest day was a round account versioning and there is not eight or nine e IPS that depend on a decision being made to to occur up to the account version EE IPS that are also trying to get in destined bull was there any direction approve or disapprove of this or was it unknown the favorite is 1702 I think I the daughter of the graph has

been changed since I looked at it last hopefully that it's still right but there are still some some questions that have to be added to the need to be answered clarified okay let's make sure that we put this as a priority to figure out before the next two weeks and just make a decision in two weeks regardless because we do need to pick one of them that sounds my opinion is except for pre compiles every other AIP hinges on this decision and will be affected by it and this is the most important change we could make an assemble at this point okay sounds good and who's championing this for this

is it just different Arthur's four different IPS its way got it do you have any comments way on this or just bring it offline to the discussion forum what was the number yep most people including the yeah P six one five team that the yet he is 7002 warrants one is probably a better option we can have at this moment for a

conversion in its suits most of our other GIP needs and also one issue a discuss is about up with both contracts I think our current conclusion is that it if if a conversion it will use slightly slightly more guys cause but it will have no issue with that so that's probably acceptable and another reason I think we may prefer were and one to burn through

is that we introduced like their sisters listo technical depth can be introduced if we want to change from burnt one if we T flavor and bumpers and the amount of Jennifer Lewis but if we T forever and troopers and the amount of trainees were managed and that's a little bit harder so yeah that's just the current general comments on a commercial so the I think the the preference right now is probably goodbye so variant one is where we add the extra field in the absence means virgin zero invariant to is where we also do variant one and add in prefix bytes to the contracts was that what the two variants

were yes so learn to like have a like defines that all the future version must have a prefix contracts of code terrific's and burn bans just say just make this prefix optional and D provides a version of a great way to opcode and contract creation collection you know determine weight so it's a the words of version of the contracts de provide for and what word one is based on either the parent contracts are always a new it version so variant one can be done and then variant 2 can be added on in

addition in a later hard work yeah yeah but I'm not sure we actually want more and two or about to go a different route because your LST several extension so you can do that can also add other like deploy use crate crate you to deploy multiple versions of the from the same account in more backward compatible way so less that may be something we will also consider but I consider the world one to be the really base layer that we can we can do right now and then you will be already used for by itself

Roger that okay great so it's seeming like we're starting to come toward a consensus on version one what are there very many outstanding questions from people where the when this discussion happen or are there any like very are there any other VIP teams that are championing an alternative to this that are very passionate not sure if these questions can be answered by the way so if they can't that's okay oh it to you or anyone [Music]

contracts and for I think the conclusions for pops a contract static account and for other type of operable contracts you can first change it to to epoxy contracts Parton and then from that point we from that point you can opt with a version of the contracts so that's less of usual we discuss and previously it's just about the yaki one six one five teams thinking it may be like having always having a prefix for account maybe visor for external tools to Louisa to understand which visa

version is a code and understand the code by seeing the general consensus right now is that we don't necessarily need to make the code you provide unction and the codes that we put into external house to be the same so for code we put into external house we can include a version somewhere as a structure but that part is just known as really need to be unchecked so I think that the to issue a discussed and yeah okay that sounds good does anyone I think we can decide this

today personally it sounds like there's been a lot of discussion on it and it's leaning toward version one heavily anybody have any objections to that or think there needs to be further discussion okay we're gonna go ahead with version one of that EIP and take note of that as something that is approved for Istanbul hooray let's go back to the agenda or was I I instantly can sleep better now that's great so we were what other AIPS are super important

to talk about from that list James Hancock or anyone else want to comment on that since we have another 20 minutes and the rest of this isn't as important those were the big ones we hit the two big decision trees yeah we were shocked all we talked about all of the decision trees that need to be addressed cool okay so let's go ahead there's the working group updates hello does anyone have any working group updates that they want to mention I only see a few people who leads that are here I know Rick's

here and I think that's the only person here who's the lead I might be wrong any any updates from you Rick or anybody else I don't think there's anything super pressing I want to discuss and and for future reference James can usually speak on my behalf if I'm not here I think the the main sort of thing that has come up recently is and it definitely came up in the discussion during this all cordobes is you know how are we gonna test these crimes Changez I don't think that the traditional style of test nets like like

the the current test nets that are out there are really appropriate because they're not gonna we want to be able to build out smaller more precise you know test nets that are still relatively large like bigger than what you look run on a single developers machine but smaller and more ephemeral than what we have with the traditional test nets and so I led to hopefully I mean I actually have to go but hopefully discussion can start somewhere about about how we should coordinate the development of those of those test environments ok excellent let's yeah let's start having

discussions on that that'll be kind of more backburner compared to the Stampley ip's but something to definitely consider well that would need to be done for is to in Pilates because if an intersection of that of Rick what Rick's talking about and testing for gas costs it's just the that all of these need to be tested before they go in and have some kind of environment to do that so that is very critical to assemble it is I was more interpreting this as certain like the fee market changes an economic change and that's why it would need some type of test set that's more ephemeral but I could be interpreting that wrong

Rick were you talking about the Istanbul the ip's or the IP one five five nine we have that requirement and you can start looking at things like well if we're gonna merge the IP 1.59 with a gas each page we would need to test both separately and then test them again together so so it's these types of situations where you're going to need to be running a lot of fairly complicated tests that interact with each other where I don't think leaving it up to the public network a public test map really makes a lot of sense test nets are more for app developers than protocol developers we don't we

have to you know we're actually changing the consensus code so the network has to shut down while we're doing that obviously right so that was sort of my thinking there ah got it take a look at white blocks open sourced testing environment and testing tools that allows you to spin up a network really quickly that has a lot of difference yes we are in conversations with them and we did look at it and I and I think the source the code has some you know there's some interesting upsides or I think the foundation with regards to that code the problem as I see it which is not maybe it's more of a concern than

a problem really is that it's a it's a SAS platform so there's a subscription model and and then there's you know devs that are the Shepherd's of that of that infrastructure that are at white walk so I don't know if that's really perfectly incentive aligned with the foundation right I mean I think there's some serious benefits the foundation to using that sort of system but it but it's not I'm not I don't I don't know how to even go through the process of determining if that's something that the foundation wants to support or if the foundation wants to roll their own infrastructure or how we want to have that discussion sure as long as it's open so the main thing that I've heard of in the past is

the fact that it's open source I know that from talking to Zack before for using it with something like the etherium foundation I think we would get some kind of credits for the SAS model similar to when we work with AWS or a juror or CloudFlare and they want to donate certain stuff about their SAS platforms or their server infrastructure or donate credits we we definitely take it up on that so we can have further discussions about that I don't see how wouldn't be how it would not be incentive aligned as long as we're not - if it helps us a lot and as long as we're not picking winners necessarily on something that is an open-source right

for me the issue is sort of when you when you pay for a SAS platform what are you paying for exactly and what sort of long-term relation I don't want to for me as a recipient right I have a fixed amount of money for a fixed amount of time so subscription service for me does it make sense and then if you think about every person who's a grant recipient doing that you know basically multiplying the cost now I'll be competing for the resources I mean that doesn't make a lot of sense and that's not really a good use of money right we would need some level of coordination

between each individual this company and you see what I'm saying yeah yeah I see that there is a DevOps team at the etherium foundation that coordinates with teams specifically and they kind of become the subject matter experts on certain tools such as ansible and terraform and deploy those things for teams to use for instance the swarm team has a ton of ansible and terraform infrastructure that the DevOps team runs so I see this as fitting in very well with that where the budget for DevOps that the etherium foundation covers this

but I'd love to take this offline with you so that I can make sure the concerns are very clearly articulated to me and then I can go back to both white block and the EF about this especially since you're a grant recipient and that is going to affect your budget and other people's budgets potentially okay one other final issue on the record with that again I'm a big fan of the white walk stuff so I think this issue would come with anyone but just you know it's also the cross team coordination right as we have patches that touch the same thing or you know for example the thing that Casey was talking about

at length earlier on you know if we've changed the gas D market with with AIP 1.59 and all that research he does come it goes out the window right so we have to you know coordinate these patches wouldn't interfere with each other on the testing side as well okay cool that is noted thank you I'm looking here and I think oh there was a note for a quick update on Blake to be by Matt Luongo and I believe Matt's in here matters we already did that yeah we did B as we learned it was a

group yeah there we go Oh perfect okay is there any client teams or research teams or otherwise that want to update anything about their clients throughout an update or announcement yeah sure so I was gonna nounce this two weeks actually a month ago but we have had eql shipping and pantheon I'm in our current updates per month it was the result of a good coin hackathon submission and it works respect so if you want to play around with EQL you can do it in the current release of Pantheon as eath quell that um a smart contract language that's like Siri

I'm sorry no you cue ball you're right I got it wrong graph QL and it's like crap grandmother we don't have youth QL grab QL superseded it was like EIP 7 some odd number but it's the graph QL spec that was all the rage at the hackathons okay perfect and I think Google also did a giant thing on that if I recall didn't fully understand it but they did something with graph QL on aetherium is there anyone else who has an update I think we actually got through this with ten minutes to spare I'm surprised just to verify it was 2024 from the account versioning that's going

through way is a 2024 version one is 7000 to 1702 yeah okay seven that was a close one [Laughter] okay if there's no other comment oh version one gonna be the version number that will be set when you create a contract now it will be like right now like 400 the legacy like like say it's existing contracts they will have version 0 and we need to define a new version for all features included and

they may be worshipped by all we can give it another random into your number otherwise ok well paint the bike shed in the ghetto challenge yeah okay okay wonderful thanks everyone oh wait I forgot the one last thing should we have the meetings at different times to facilitate people like Nick Johnson and others in on that the part of the world that's like Asia and New Zealand and Australia that it's like the middle of the night for them what do people think of that I would support it I kind of brought that idea of the idea

is that there would be an eight-hour shift for each meeting so that you know I would be meeting in the morning and then I'd be meeting in the evenings and I would need to wake up in the middle of the night so that way each major timezone we'd only have to wake up in the middle of night once every month and a half and for two-thirds of the time they have a rational daylight time to college these meetings okay that sounds like a pretty good plan and that would happen on we would still have that on Fridays then that would have to be on Thursday because I mean Friday is for some people and Thursday for her for others that's right yeah so I think what would be +8 from the current time and

be minus eight from the current time whatever it is yeah that will have to be minus 800 s Europe ends up but no in in the night so yeah it's going to be evening in Australia morning in Europe and evening the day before in the Americas ah it would be Thursday evening and the Americas and then Berlin it would be morning and then Asia it would be afternoon I get that right at least that's the way I understand it so that everybody gets included okay and the thing I wanted to avoid if possible is like people like to do stuff Friday evenings because it's the weekend so is

there a time shift that we can do that gets us into like Thursday / Friday did we just describe that yeah that sounds good to me um - 8 - 16 uh yeah let's do something like that I don't think that's gonna I I think that the e 2.0 calls are on alternating weeks from us so they're never in the same week as ours so that would be fine mm-hmm okay let's do that anyone opposed just to make sure innocent so minus 8 from the current time minus 16 from the current time

yeah oh my god exactly what are these cries that would be exactly what it is yes okay yeah - okay cool okay I will announce that in the get er chat and start putting down for my personal calendar what shifts those are every two weeks so minus 8 and minus 16 from the current time of 9 a.m. my time okay I'm sure Nick will be happy about that as well as others in the Asia timezone

who might want to participate um anybody else have anything yeah just a quick comment yeah I was talking with Zach right I'm from white block I mentioned what Rick had said and he he wanted me to just remind everybody that their stuff is open source so people are free to use it like that and he's open to talking about like how they can support the people working on yet because they need testing that maybe might not be able to do a complete software purchase or SAS purchase cool now that's nice of them and thanks for bringing that up with Zack Trent okay if

there's nothing else then we will see you in two weeks - oh wait oh we're doing - a tour - sixteen first doesn't matter pick one up to you oh I'm the decider preference because - ooh - sixteen is gonna put Europe at 11 p.m. - 1 a.m. so I'm not sure you'll get a lot of seekers and there's a lot of coordination in Europe yeah let's take - take because we need Martin to call in ok - 8 it is all right thanks everybody we'll see you in two weeks

good bye here we did it hooray [Music] [Music]

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

[Music] [Music]

[Music] [Music] [Music]

[Music]