All Core Devs - Execution #22

Transcript

um they're kind of I think it's being from Korea that's what they're hating on yeah I'll have to see I'll post the link in the in the scroll box I mean in the I'll post the troll box link and github is what I meant alright alright well I'm starting yeah okay let's check this out alright first agenda item thank you everybody the subtleties we're gonna work out let me see here is our kitty on hmm I don't see this okay in that case

we can we'll just keep going for now this is the 649 metropolis difficulty bomb delay in issuance reduction I think that a furry was going to redo that let me see what his updates were yeah I see that okay it looks like he just did put in the changes that we talked about last time which is to reduce the difficulty bomb as Vitalik had said there's still some there's still a little bit of discussion on that but I think we're generally set it was specifically this comment I'll put it in chat

so what this is trying to do is this is trying to replace 186 which was going to where is that oh yeah reduce issuance before proof of steak so I think all this did is combined it yeah so yeah that this looks good to me is anyone have any comments on this hold on what specific comments was this was this the one I wrote or is this a different one the one I pasted was the one you wrote that I believe every included in here but I'm double checking right now okay yeah yeah and I guess the main question is was there anybody here is the link to

the file I I don't think that this has any difference for okay yeah the block uncle and nephew reward rates was three trillion if block number is greater than or equal to Metro four of three three three actually so the new uncle reward is eight minus K times new block for Ward divided by 8 and the new nephew reward his new block reward divided by 32 this looks the same yeah okay like I mean like the simplest way to define it would be just like a across the board

40% reduction in all of the balance changes in the blog finalization function but I think I think everyone understands what what the spec is okay sounds good so any other comments that need to happen on that feel free to go ahead and do that mm-hmm okay which one away aiming for with this one yeah oh this one is this something 649 well yep I say that one more time Martin I think I missed it which one of the ports the first or the second 649 is oh yeah that's a good question

or I think it was the first is what we talked about what you mean yes it has to go into the first because like there's its if we wait until the seconds to do the dollar the issue it the issue its reduction anne's the block in the ice age delay then the blocks are just gonna be way too slow right mm-hmm okay cool so for the next item unless it was there any other comments on that it looks good to me but that's something that can be quickly that can be kind of finalized next meeting if need be but I don't think we've changed anything about

it in the last few meetings how do we have so why do we think we have an update on the projected block times um I can do that right now meanwhile some is this is this finalized one 649 yeah I wouldn't say so because a free just made changes on it like three days ago but I mean it's for the for the purposes of this call I was looking for anybody who has skimmed it because it's a real shorty IP and found any problems with it but I wouldn't declare it finalized to the extent that

we declare the IP is finalized I think it sorry so I got the so here are the results for block time increases on August 25th we can expect an increase to about 25 seconds then on a September 24th we can expect an increase to 32 seconds and on Halloween so October 31st we can expect it increased to 40 81 seconds okay cool let's see so there any

concerns around those numbers being given whenever I did kind of a really unofficial reddit thread about the block time everyone was kind of saying after 30 sec or after 40 seconds it would be a little bit tough but right now there's no there's not many gaps in production that require a sub forty second block time yeah so one thing we could do is try as an initial like to see if we can get metropolis in by Oh September 24th which would be four point three mil though I'm not sure if that's like if it over-optimistic over-optimistic not optimistic then yeah I think that we'll

have to get to the point where we talk about testing and which ii ip's are going in bread because i think there was some confusion you know so let's go ahead now that the block times let's reappropriation [Music] for either accepted or under consideration absolutely I think there's a pull request from I thought there was a pull request from a free to add that so I'll double-check that and if not the mere case he'll change it yeah it's good those collect them definitely okay so

the next item should we defer AIP 96k see you can go ahead with that on the agenda page there's a link I'll paste it into chat as well that is talking about he I P 96 and why we should defer KC wrote up as a thing yeah just looking at the progress made on different ones it seems like 96 legs behind a little bit I don't need to repeat everything I wrote in the comment but good wanna hear what other people thought I mean I definitely be okay with

delaying 96 if it meant we could get in the trouble we could get Byzantium out faster yeah I'm looking through this my question is does this affect any of the other AIPS that would be going in or can it does it stand on its own it's totally standalone okay then yeah I think that as far as you know getting the difficulty block de life change in there I think that that could be a more important yeah it sort of depends on whether it's more difficult to pull the change out of the code bases I already

have it which i think is mainly C++ and parodied um first of all it's not so much about holding it all we have to do is just like a rename it you know change the word Istanbul to the word Constantinople and then set the Constantinople for walking over to like Texas really in okay I see what you did there all right calling not snap helpful instead of instant bull is a rather political move hello I think in case of this the AP there are some pending issues there and and I would say is

quite a lot of unanswered comments about all of the changes it's into some details I also put a put a comment in the Indiana just after Casey listing some of both points that are coming from my side but I've seen there are some you issues issues as well not clearly answered I think so I would say the pace on on this one it's not ideal so I would also go for the languages to the next hard work okay yeah is there anyone else in

agreement on that I've hung it up good okay that's pretty much everybody so I think that's good that we delay that so unfortunately we don't have an imperative folks here yeah exactly I mean and well-well say for the moment this is okay and I'll double-check with Arkadiy cuz I think I'm gonna be talking to rob soon specifically so next item and I'm actually actively updating this but let's go to the one that Tim posted I'm gonna it's a recent comment I'll

post in the chat and then get her so this is it it's basically asking about the big aunt Maude expe IP 198 the requirements for the benchmarks on that so Tim you can go ahead yeah so yeah it looks like here up here up here just commented on that and I said it was mostly finished and thus he was gonna maybe add some some other requirements later but yeah I'm not sure what you guys wanted or needed in regards to that specific issue so for them modok's

I think the there was a discussion if if short inputs were penalized unfairly I think that's been settled and if mr. long what formula to use for the molex well it's not really a local from Alex and the rest are the specific costs but as far as I know that formulas or agreed upon cool cool yeah yeah let me know if anybody needs any more benchmarks or whatever yeah thanks to him it's been helping a lot I've noticed on the

testing channel so that's great and them and does that pretty much wrap up that point yeah more or less I'm going to try to set up like some kind of framework or check out holy mins uh testing framework yeah that's it thanks great just a quick quick mention regarding one of the pecan past mod X yesterday I was working on the merging all the pre compiles into get and I double-checked with the other in foundations just out of curiosity and I saw that parity is still using the old original gas calculation and I tried to reach Arkady I couldn't so if anyone here speaks authority people please being them that the gas calculations

changed yeah but Peter we discussed this in the record of channel a couple of weeks ago and I got thumbs up from I think Robert Robert and unpaired team so wish I mean it should be aware but of course it's good mix next aware okay any other comments on that okay great so it looks like the last point for Section

A for subtleties or questions to work out is me I'm gonna ask as a IP eighty-six the only thing going into Constantinople which is the hard fork part two so it sounds like they always know exactly and then so the only thing is we talked last meeting considering the bitwise shifting IP 215 which is the one that Alex Bears Aussie accent brought up the link to that's going to be an item three of the agenda that was shifting to replace the IP 145 so is that something to still be considered of enough importance to consider putting it into Constantinople since that was brought up I'd say it's

worth considering but I don't think it's that useful to spends discussion time on it until basically the every metropolis has settled around the road to being settled okay Alex do you have any comment on that so the first half was breaking up with you he said Hudson oh I was saying that um I was asking you know which ones are going into hard fork part to Constantinople and for that one you said you made a comment recently that he I P 215 bitwise shifting was considered as

part of that and since I believe that's your EIP what's your perspective on putting it into metropolis at all well of course I would be happy but it's it's really up to the people implement you know their different clients and the client I could handle is the JavaScript and perhaps the C++ and I would be happy to work on those two but any other clients it's not really up to me and but I would be happy to do you have it sooner than later of course um I guess it's not such a high priority at the moment compared to the other work in progress understood yeah I would agree there and

I did I just think it's pretty cool anybody else on the line have an opinion up or down about this okay yeah for the moment I think that we don't want to put any more on our plate it's kind of the the feeling I've been getting from other people so we'll just keep this off of them the metropolis schedule for now so let's see if there's no other questions on that let's go ahead and do updates on let's do that actually instead of updates to testing

let's do updates from the client teams first death is there any updates on your end worth mentioning yes sure so felix has been working with the C++ team on test generations we are as far as I know all the tests that have been generated are currently imported into our repo so we have all the wrappers I think for all of them and we also finished more or less polishing all the pre compiles and basically just ready for merge so whenever marking our Felix sums it up it can be added and the last batch of polishes that we need are the the opcode

VIPs in theory they are done in practice I just need to polish it up and merge so [Music] being one of the few issues that I do have is that for example we've merged in already way a few weeks ago to EITS one of them was the difficulty adjustment which obviously needs a rework due to the ice age and the other one was removal at the intermediate state route which now was obsoleted so it from my perspective it would be nice to have a list of the IPS that we are finalizing so that we have an idea of where we're standing because it's a bit annoying

that that we have to weed out previous e IPS that were at least three vesicles finalized so we can walk through the github comm so I should the REM /e IP is a repo I'll just link that again so if you look through the ones that are in there now number 90 six should probably be put into deferred because we're moved in to Constantinople then 98 should probably be just removed entirely because 658 supersedes it then 100 is still fully intended to be happening then 140 is in full gonna happen 196 is gonna happen when ie

sevens gonna happen 198 is gonna happen to all I was gonna happen supports an 214 is gonna happen and we also needs to add into that 66 69 that's my understanding yeah it's a 649 / 669 yep okay that's nice big yeah and we can update that list after this meeting cool so that was death do we have anyone here who can speak for parity Mathew English are you with a client team or consensus or okay

perfect just making sure I thought you might be worth paradise actually I've been working with death suite welcome um so let's see let's see parody well I think last time we checked they said they had everything implemented as far as test passing does anyone have has anyone seen them comment about how many tests for passing versus failing recently Martin I know you've been talking to our kitty some yeah I've been quite out of the loop for the last two weeks so I don't I don't know the hive tests show a lot of failures and current don't think they're running I need to

look into to that I'll know more in a few days sorry oh no problem at all okay so the next one would be C++ aetherium which I think would be with Chris Powell yeah yeah I can I can start here I think there were there were no functional changes in terms of yep implementation Demitri it usually is or hard working on tweaking that the test let's suit and bringing more tests to a piece so that

was the main the main point of full following hard folks implementations I know if you if she wants to add something to this now we are working on now we are working on our uncompressed generation the crazy studying code coverage approach from different clients what I want to do is fast testing tools that could run a random test on every client and I bet they need support from github go I mean from other clients I need the comment that I could use to

execute this randomly generated state test at least microwave yeah I think we've come quite far both on get party and CBP Empire areum and yeah I think it's just one thing that it's relative parity and that's the receiver come on we'll just have to bug them a bit more about that so I think the way to proceed here is the first the Demitri provides some profile and maybe in Demitri can also create issues on each client Beppo the file is the same as a state test and just the state test but it's

randomly generated it has some random VM quality if you have finished implementation or not the in order to know that in it's a it's convenient to have a fixed man pie-ya any test from the state tests folder on repository and just and I need a comment from like get that I could provide this pass to a test file maybe is to the input and then get could run this particular one file is one test you can already create an issue them because

you know I mean the commander should take this argument and does this my idea will be to to have put up a server which does 24/7 fast testing on all kinds which can be fast and then the outputs from that would be consensus failures where you could download what the command big traces and the commands needed to Genesis and the code needed to actually run the EVM and reproduce the issue that's kind of what I'm hoping to put together my same as Casey yes actually isn't our isn't the EVM

utility in GUI theory I'm suitable for this test exactly yes that's what I've been trying to achieve with a EVM utility and the some of the part even been which I think it's Thomas has been done most most of the work on that one the patience every VM utility doesn't calculate the first state hash no it doesn't but what you can do it get gives to the internals free job code and that's kind of enough for it to do equivalence testing it also needs to a few tweaks to calculate and prepay the gas and and charge for the intrinsic gaps of a transaction because when it

when the state test run all that stuff is already done but with the EVM command currently that's not it doesn't do that yeah yeah so maybe some additional tweaking is needed let's have a chat next or the next week about how how we need to tweak the clients and the events to get what they want okay any other comments on that it sounds like a separate meetings being set up cool lastly okay yellow paper yo Ichi I'm

guessing there's not really much update still there's a pending task of separating metro police in the prison gym and Constantinople that has not in here some coordination with giving is necessary I think okay sounds good let's see the next one is pythonpath up Vitalik i did run the tests for metropolis and they were there they all

seem to be passing the there was one exception that has to do with code size and I believe so we still needs to confirm whether or not the code size is 24,000 bytes or 24,000 576 so that was one thing that we needs to confirm and so I vote so I'm pretty sure that it is 24 M exactly 24,000 is I believe the most good ions implemented the other number the other one I think so okay so this is importance can we quickly check right now to see what client what number is implemented in particularly in depth

and parity yeah we can looking up the hive oh if there is so what's the name of the Brooke chain test and then you know I can look up the hive result for that I mean the simple thing is I just go through the source code yeah yeah maximum capacity yeah can someone go through parities who's most familiar with it or if the hive tests work for a quick look that could do it to everyone else hears from Gath mostly the gut code size max code size is 24k 576 okay

I'm just the clarify those are what a contract of size exactly two four five seven six be accepted let me check yes okay yeah let's check parity to Nani the max school size is accepted cool is there anyone in the call look at parity or I've barely heard it off of 24 576 and is 24 576 exactly accepted so it's 600 in hacks

maybe so let's jump yeah so thank you for about 7 seconds on the fly that is 2 4 5 7 6 ok so this is important because the EIP that introduced this change number 170 actually said that the Mac says that the maximum code size is 2 3 999 and somehow we ended up diverging in the VIP I'm also yuuichi can you check what the yellow paper says so first I did this 24k and after Y was anticipated and I asked around and the people told me no no no don't look at first comment in de

patatas to regret something video that's the right number ok I don't remember who told me but I thought that so I'm sure it's a nice X number ok um let's see parity max code size ok write a check for that link and then ok so that gives you max code sides and now we want to search through how the number of merit max code size actually gets used wait hold on mmm I'm not seeing this okay so

can we spin you search for the line of code in the parity coins we're back to code size actually gets called or I'm trying to do this and I can only finds the JSON files same here yeah well they put it in a variable with underscores instead open Keller it's tight on the score let it come to create the element right okay if or data 1 is greater than okay

again already it accepts exact size if you look at eath course wash as a mess or sees a bank there in the Tuck yes there we go it's a line 311 of this file - um okay yep no that's okay good that we've cleared that up the other thing I wanted to confirm is our reconfirming like in the in the Skype channel uh yeah I I had suggested the change that we the change retroactively from the from Genesis that if you try to create a contract at the point and the contracts code or nonsense non empty then

contracts creation fails immediately so before initialization even starts and my reasoning for this is that number one this is ultimately the basically the only right thing to do once EAP 86 gets implemented and number two is that it's the only way to trigger any of this before Constantinople is basically to do an address coalition and there's plenty of worse things that you can do with address coalition's anyway Israeli is there any thing that is pressing that we do this now instead of um there's nothing pressing I guess oh we could if

we wanted to we could just leave this issue until Constantinople but basically there is one test which is failing and it would be nice to a you agree on what what should happen in that test I mean if this does mean just not thinking about the sons of Constantinople and that's fine too because as I said the chance this'll turn into a consensus failures microscopic so for example I know for a fact that the when it turns out that the simple process implementation of this well we should I figured out the CEPA specification of this was part so if someone can create an address coition

then they'll create consensus failure you have this change will make the implementations a little bit easier because just because address collisions can now override code the generating system needs to keep track of all the codes so that all the codes can be recovered but that's only necessary for address collisions and if this proposed change comes in we can make the general system already easier yeah I guess is

good change I mean because if it's already consensus issue it's just very pretty fine the definition in the edo paper now is that religions not considered and when the contract women contract is deployed when the contract is created it starts with the empty code so address collision overrides our contract and do it today have tests already which so yes we did yeah so it's just something that needs to be implemented in as well succeed so

sem / tests already have this test so basically there are two ways I mean yes didn't know that the for the way would be just believed a yellow paper and try to get it right and she propecia PP theorem and Cassius and oh brother crayons implement this for this virtually never happening case the other way is the Daleks proposal yeah I would personally go with divot

Alex don't belong here it seems safer and more aligned with systems like get and others when you never accept the same harsh to be downloaded okay Vitalik did that cover that topic well so what's the this decision for a moment I'm not sure there is one so if I hear no positions I'm will go ahead to

changing the CPB theorem change the test Suites with this respect does it record if we're going to disallow overwriting existing contract does it doesn't need a VIP for this to be specified correctly or some test and force everyone to be compliant with the test yeah so my current my opinion is that the probability that any of this is going to become and that any of this is going to be an issue at all what even in the presence of attackers before

Constantinople is very small and so I would be personally view it as being kind of similar to Yankee 170 where we had implemented the code chain of the maximum code size limit and implemented it retroactively Lee Webb least we may have employed to retroactively wave forget it so I you know I'd be happy to write any or write-up in Yankee for this if you're quickly if we want to do that for procedural reasons that's probably good because there's no one from the parity team here and we're missing a few other client teams like Java so hmm yeah

thanks italic let's see so did that cover that I think it was the talaq who brought up that point initially okay great and we were in the middle of seeing where paya Thap was so paya Thap you were just saying you were figuring out that main point oh your cousin I think we should be um can you hear me now yep I can hear you now okay what's those two points address we should be fine on the state tests but I do want to call attention to the facts that the block tests the the block specific IP is so specifically 9 th in 106 69 them haven't

been implemented in PI theorem yet and basically I'm just waiting for tests for those to exist east up I am going and running broke train tastes yes again hmm okay okay I'm to do tests exist for you for any of those um actually though the pie here I forget though the pie theory of coins can run watching tests but it has not been passing them for a while and part of the reason why is basically confusion on what box schedules are supposed to apply to what sets of block tests but I mean we can take that offline and figure

and figure that out I do the block chain test currently implements in the apakah block specifically IPS transaction this sheets yes okay and specifically the version that replaces the transaction I received with a single bite for success or failure rate yes okay what about 90 okay so that covers 98 and 100 the uncle ins difficulty change is the difficulty in the broke chain tests should be already computed with the new formula and I had up to here with some tests and I changed it no

when somebody changed the difficulty formula so there are there are tests involving unquote so there should be okay okay so and here's some VIP is 669 m oh that's a 10 CPP aetherium now yes but in that case okay then Oh we'll take this discussion offline and we'll fig and figure out how they on all run pi here and throw the watch in tests and try and talk it talks for any of the ones that yes I'm interested this is a priority if there is a test amazing okay um sounds good to me in that case I'm I'm satisfied yeah okay

great any other teams in here Tim are you mostly doing Gotham Gotham parody testing or you with a client team it's a good question i bouncing all over the place between gas parity and the spying on looking at the c++ swing tation pasta i i at that for whatever else yeah cool well hey thanks for thanks for helping out and so you're not on the etherium j team then i'll try to get one of them involved to see where they're at not that they're a client that has to be up to date on metro necessarily but i think that's what they were potentially aiming

for see I don't mind looking on either that's for Java yeah a theory I'm J they're getting back up and going again for Metro compatibility is my understanding okay yeah cool all right so that since that's all the clients I think with three office we've been talking about testing but Demetri Martin are you Ichi or anyone else is there any other testing updates we should be aware of and is there any type of percentage or quantitative measure about how close we are to being done we have a splitted the test with metropolis field to Byzantine and Constantine oval and there were some

updates to block chain tests nothing that often new features but just cosmetic updates and Vitalik you said that you don't know the schedule over blocks and blocks in distance now has a network field that's specifying which rules to use when computing the death and then there are transition blockchain tests specifically to test them transition from one program to another and the block five and we could add okay is there a document like that it's written out that describes how all of these rules work they're pretty speaks for themselves

likes the transition block five from this okay I mean I recommend it's like having that kind of explicit document written out but as I said we can go wider I pasted in a link from from our internal test suite which initializes what might be a hint might might help you yeah and also hive tests are generated from the state test they are also in a test repository and develop please check always check that here on latest develop of the test repository and you could see on the blockchain test there is a folder general state test which actually abroad scientists representing them and this are the tests that I should personal clients so

Dimitri I'm curious earlier there was something called transition that with transitioned at 5 8 10 14 and 200 I sent link there is that deprecated it's it's link that the Petersons better source of truth and link right now as I ended in the chat Google hangout yeah yeah it's a link to the go source code yeah I thought about that rules for transition and yam this is deprecated now

alright so update single in the transition okay so if I understood you right Dimitri you said all tests are written right yes deprecated well to solve that problem I I want to run them random test generation like that cases country working his own approach me I

never see generating a random founded called and random scenarios of of code combination from other clients and see maybe yeah Casey was telling me about the first testing it's really really awesome looking so if anyone hasn't talked to Casey about that or seen it talk to Casey it's really really cool yeah okay case she's using some tool with colors all of that possible scenario store called carriage and then he gives us for one single client the sheep will be able to on this test and all other clients and maybe we could find solutions we are working more on coverage with the fast and random

testing the the art of first testing there's actually a whole subreddit about first testing by the way well but the Hudsons question we have measurement or both progress how many dissipating of æneas exciting um when the undying hive statistics works it will be that measurement currently it's weird because in epidemics - tests metro police case have disappeared now there are only

Byzantium tests and constantinople tests but on in the undying hive interface that we just see the metropolis test yet so the measurement is not up to date so yeah when this starts working we can you know for all the numbers hopefully decreasing yeah I wasn't however it hasn't been updated to these new networks these new rule names that's why we're not seeing an emphasis on height but it's already helping a lot okay that sounds great

so that kind of goes over testing so we're back kind of to when the testing and release should happen and Vitalik you remind me when the what the block time when it hits like 35 seconds did you say um hold on let me look through that again okay so September 24th is when it goes up to 32 seconds that's right and hello and Halloween is when he goes up to 41 okay is it feasible to get this in by the end of September so that would mean

by later this month starting testing for I've heard numbers being thrown around for amount of time to test and I think that we hit a sweet spot on four to five weeks right yeah and so I think right now the major progress we've had with this meeting is that it seems like there are client we've established or like we established a full specification of all the e IPs and it does not seem like we're arguing over yet here protocol anymore and that we have clients that have implemented them and it seems like we may already have tests for all of them so the next stage at this point seems like getting to the

point where all the clients are passing all the tests and we're comfortable were just enough in terms of tests there are two things to catch up 96 should move to Constantinople and this hasn't been done in the tests yet so this has to be done this ice age DJing is not implemented so this has to be implemented in Chicopee artists yeah I would feel good and I just want to see everyone's opinion on this if we can get some approximate dates that we think we would be able to launch testing today just because the next poor devil eating is August 25th and then the one after

that is September 8th and then we're getting into some higher block time territories so I don't know for instance like which teams have the resources to make sure this is done by then at Martin what was your opinion well it's gonna say that I mean testing is happening all the time but I think that we should perhaps start weekly calls people who are doing testing and our writing tests and doing these things to really get that ball get all the tasks done that needs to be done testing phase could

also mean activating it on the the testament on on Robson yeah and actually with that would Rob's to be the network we use and is it stable enough well that's the only network at all science peak so I guess yeah unless we just spin up a whole new one but yeah that I think if you like the easier thing is to just do Rob stone well it makes sense to activate it independently on ANCOVA and rinkeby as well yeah Peter what would you think about updating rinkeby with with this type of thing basically that's just modifying a

configuration so it's for me it's all the thing okay cool all right anyone want to be brave enough and throw out at a time we should be doing all this or should we wait till next time they declare a time I keep getting worried when we push it back but if I'm the only one then I'm cool waiting would it feel meaningful targets so let's say that let's try to wrap up and try to pass all the tests by the next quarter meeting okay yeah that would be okay so if we're doing that if we're wrapping up and have the test done by the all core dev meeting then we could feasibly then like something like

the last week of August start the testing period right for a test net or the week of September yeah first week well basically a last week of August August first week of September kind of period yeah the talaq did you I thought you were about to say something I didn't see no but no I was fine to me okay so right now we're gonna shoot for testing the start somewhere between the last week of August and the first full week of September and then testing would occur for do we say what was the number for weeks or actually can it be shorter than

that or I mean I feel like we have this discussion and multiple chord to have meetings ago but there's like some there's not like basically the more weeks of testing doesn't necessarily provide that much value as you go up a number of weeks past a certain week do we have any features which are already entirely finished everywhere and could be launched already or do you once you launch everything at the same time I think everything at the same time just like how it would be on live and that's usually how we do it right yeah but there isn't really a better way on

this we want to simulate three small Forks yeah I think that we should just do everything on test and at the same time everything on live net at the same time just because we're already breaking up the hard fork to put things on a second part of it or a second mini hard fork I guess we can think of it as so breaking it up for the actual launching of the features further doesn't make sense to me but I'm not actually implementing these features so I'm totally happy to hear anybody who thinks differently I an easy way to launch one feature but not launch the other ones so

it's gonna be just messin okay okay so yeah we'll shoot for end of uh yeah last week of August first we have September and then four weeks from them so we'll basically be getting it done by let's see one two three if we do four weeks of testing it would be first week of October if we do three weeks of testing it'll be the last week of September of course depending on when we actually you know launch it on Rob stone oh also I'm guessing it's better to like just just as we're deciding dates and stuff as it looks like it's better like during the

week to launch these things right cuz that's like Work Week instead of a weekend yes okay cool I'm with that too oh man my birthday's on September 24th we could it's a Sunday though so let's not launch it then it's about to say I could get a birthday present but yeah that'd be a stressful birthday present let's not do it that day while we still current target 44 the 41 second mark has some nice candy and ghosts on it yeah where else am calm and Cancun them oh yeah oh yeah I feel like I feel like multiple teams and

multiple people would have heart attacks if we did the fork during Kin Kin so yeah let's not do that okay cool so we have we have dates at least for testing done last week of August first week of September time period so yeah I have something on here for gas prices and opcodes for pre compiles Martin I think we've been going over updates as we've been talking but is there anything and the updates you wanted described or anything any needs you have no I haven't been able to spend enough time on this but what I can say that it appears to me that the new of codes compared or new precompiled compared to the old ones it seems like the ones are too cheap and

the new ones are a bit too defensive too too expensive and should be made a bit cheaper and perhaps the other ones they're not in sync and it looks like Tim and Matthew had been messing around with running some of those on different clients so we could probably get their support a little bit more if needed I think yeah I think I've been seeing that in the testing channel at least so feel free to to ask for help as you've already been doing yeah I'm back and working closer now so I'll talk with Arkady and the other guys for doing benchmark yuuichi and Tim and we'll try

to sweet decide on something okay cool we already talked about that why shifting is there anybody else who has any other topics that aren't covered on the agenda in this call nothing okay I'm gonna post something funny I found in the troll box basically someone kept seeing Greg's picture and making up like different scenarios about Greg being this wealthy wizard with birds and his beard so you can you can read through that if you want wealthy you've got to be churning yeah apparently you own

likes 80% of the theory amande like everyone way he was like I want to keep that guy tongue I want to keep hearing that guy talk and it was like all this other stuff so yeah just getting just gonna end on a comedic doubt there all right cool there's wood grain yeah if you have any wisdom to share Greg before we go what I don't haven't raised him at all okay so that guy them I wouldn't be living in a beat-up mobile home in the middle of nowhere [Laughter] well thanks everybody meet a year ago okay thanks everybody have a good day

we'll see you two weeks