All Core Devs - Execution #49

Transcript

[Music] [Music]

[Music] [Music] [Music]

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

hello everyone this is the meeting of the etherium cabal and all who are included in the secret rituals we now are live-streaming that as we have for the past two years let's go ahead and get to the agenda is anyone here who can kind of discuss testing I think Martin maybe yes so testing I can't read it I don't know exactly manual just but the first testing has continued and it's become a lot more stable so right now we

have two machines up with different coughing strategies and it's doing like it's done five five million test cases it's October that's taking along two machines last 13 days there is it has found one issue one consensus issue client which is being fixed yeah okay great

yeah pardon the dog in the background if y'all can hear that I'm at a friend's house that has dogs so let's see what's next on the agenda do I close out of it now I'm here so next up we have the client updates so we'll start with parody if offeree or Fred want to jump in sure we for cove on to the latest changes and seems to be working we got on that issue that was reported during DEFCON on block propagation or working on it or also talking to the

lights block I believe their name is the simulation tool to try to see if we can get some simulations on what we're changing this means but yeah looking to get out of fix it otherwise no major updates okay guess hail so I guess we have basically the same boat we just have a whole lot of bukas piled up during descon and the previous week's for us but everybody was preparing so but just kind of crunching through those nothing

spectacular okay do we have anyone from aetherium jay here yeah it's me McHale hey yeah nothing special here as well preparing their release with fixes for issues that has been have been discovered during Robson hard work yep and continuing to work and it's 2.0 implementation yeah that's okay thank you and Peter could you meet oh yeah sorry

no problem c+ r LS just slowly preparing which will be and besides that DMC points okay aetherium Jas yes it's me hi yeah we made good progress on consider

no we just merged the create to a code and now it's just a Scottish which has been open and we're just waiting for the tests and I think that just just got merge today in those tests or at least some so I'm confident that we can do a release in one or two weeks on on Constantinople great that sounds awesome next we'll do you Nimbus like everyone no real updates just going out there calm stuff yeah I think we're all recovering turbo gas hi so yes I've been recovering but also I've started to work

again and that issue they had the bug that I discovered during that they wrote in Hartford it was something to do with rewinding and apart from that I been working on these molars database that have announced they're four and four and also I started to try to extract some data sets which I will talk about maybe in the next meeting or something which is de trying my hand at using the efficiency of turbo gas in extracting all different types of data from the database which is a so far we're looking pretty very good so yeah that's it for me thank you Pantheon yeah we just had a offset all

week in Prague I'm actually in the airport right now um we shipped our Pantheon release and it is available on github Pegasus edge Pegasus at the y en g / Pantheon so the pull requests are open and issue queues are open and we also have hope you just muted Anna hello Danna okay I can't press space when I'm on the other page on my Mac so we're gonna be contributing some reference tests back they're probably gonna be working on the ext you got me - - yeah I can't keep in spacebar press and switch

for sweet pages but we're going back to the XD code hash we're gonna contribute back to some of those reference tests so that's the update awesome let's see Trinity roughly the same we are I believe entirely in theory constant no ready but we still have to get all of the urea tests tests passing I think let's go a few failures there so working on that okay can I add a small thing I just realized so we're shipping an alpha it's just an alpha is just for testing it's like calling to wallets today so if people want to be on the bleeding edge and try

some stuff out and see if they can break it checkouts feather fees Google parity feather I'm sure it Joseph great Who am I missing I think I got most people what's your client am I missing okay I think I got everyone then that's good so next we have research updates I don't know if anyone from the research team is on actually so he might skip that anyone here who can represent research or any

part of it obviously not speaking for research but Hudson do you want to do he wasn't as well I still know but yeah the respondent under clients or research ID if I put it under research just cuz okay so I guess we don't have anyone from research research then right yeah go ahead with he was oh cool yes so we we publicly launched our test at at DEFCON which we're pretty excited about so that's live now if you plug in you was not a theory a morgue with I think the regular RPC port which is eight five four five intellect meta mask RPC etc that's running that's only on guess right now there's still a few

outstanding changes that are pending I think updates to EB MC etc so I think we still need to publish the latest Genesis data into our like public repo or whatever to make it easier for other people to add nodes and we'll keep publishing updates into Gator Gator - I am / you awesome / lobby and I guess the other big thing is we're still working on something outstanding design questions like like a big interesting one for example is like static and dynamic linking like how do you link multiple it wasn't contracts to each other and I guess that's what the high towards hopefully maybe having he wasn't on the main that sometime in 2019 although that path is still not too

super clear I think Jared oh he was here I don't see him now it's gonna ask if you had anything to add Pavel do you have anything had oh I think Powell doesn't have audio anymore looks like it dropped off sounds good alright next up we have Constantinople we have a few things about this before we do that let's talk about the Gourley po8 s net

because that might be somewhat included in the discussion about Constantinople whether or not we're gonna use that as one of the test nets that we can test Constantinople on so offering if you can kind of just give an overview of Gourley and also tell me how to pronounce it correctly that'd be great yes so girly was born at the East Berlin hackathon as an idea and initially we wanted to have like a cross cross client proof of a sorority Network yeah and now with the current status is that we have two people that stepped up

to implement click proof of a sorority engine to to parity and also with last week's release of Pegasus we have like let's say two and a half clients that support click and idea when I came back from Def Con that I just we started early that girly that's how we pronounce it was was pantheon and yes many data that pretty smooth and I have to admit I'm surprised house amazingly simple clickers and just works and now we are in a stage of a free chest net for the

GUI test net and I'm basically waiting for for the code for parity to be ready however ot allows a mode to skip seal validation and right now it's possibly was synchronized parity Shane the girly Shane was a parity client in an insecure way so far we have like two dating clients and girly and one not really validating triumph yeah ultimate goal is obviously to have for chestnut for all clients and to answer your question for Constantinople I think purely it's not

ready yet we are not there and I think purely regarding the code map will be probably ready after Constantinople that's why I'm also activated all Constantinople changes in early Genesis so yeah it's a fully working Constantinople chestnut once clients have yeah any question just to add a bit I'm not sure that the you saw that currently it seems that gorilla does not this seems to be stuck I think all clients for the oil yeah I didn't check today but last time I checked yesterday we have our first consensus

issue yeah it's a long way to go I'm I was quite excited to be able to sing like four thousand blocks with sweet often client on a few I network but here we need some work and yeah if you need any help from our part just chart for client developers who would like to add support for that where is a good place to start oh so we have to get a blending page it's girly with o e so the things people are complaining about this this letter with two dots always to dot the German ooh and if in case you don't have it or like the

computer signs representation for that is like Oh II so you just type girly was oh ye and go on github early test net and we have kita Shanna was the same guitar dot iron slash girly slash test net and both tests are good to get started contribute and shout out to all people that are already contributing it's amazing how many people support this idea around this initiative thanks so far if I could just chime in really quickly I'm working on the click

implementation for parody and excuse me I'm currently technically on vacation but I'm hoping to make some pretty decent progress over the next few weeks or so really excited yes thanks for doing that I'm on vacation yeah had to come on vacation to aetherium doesn't stop and we have a problem we can't stop working so basically I have a couple questions so the two-and-a-half clients that have

it working you said that's priority get and Pantheon is the half I think ah I see okay and then the other question is is click the one that rinkeby is based on yes okay great those are my only two questions well that's exciting so now we can jump into Constantinople there were if you go to the agenda there were two comments in the agenda that kind of suggest certain things and it looks like there was a suggestion so a free suggestion was to aim for January

16th which would be around block 7 million 83,000 314 for the hard fork so and that's just putting a number up there for some base discussion and then ogre has a more I guess complete plan that has to do with starting a new proof-of-work test net and deprecating rob stand on December 10th forking a new test set on January 16th and then fork on main net February 12th so that one's a little more spread out let's go ahead and go with Oliver cuz it sounds like he's been thinking on this a little bit did you want to comment kind of on your

plan and hear what you're thinking on this um yeah some comments but no that's not not really a plan and I didn't really take on this and it was I'm just coming from a software engineering perspective and from my general intuition this this I think this would be a good plan but are a bit more conservative planned but I'm for example no one who's setup chestnuts or something like that and I can't really judge how much effort it is to have a new chestnut and also not a heavy user of the rocks and chests and so and many open questions there it was just one suggest because I thought it would be good one alternative to this other

suggestion just some discussion ground but I haven't put very much effort into pushing this further or something got it ok and we had discussed the test net last meeting and if I recall someone correct me if I'm wrong we kind of said that there is some work to start a new test net and we also like the fact that number one rob stone is heavily used because that gives us a lot of data and like a lot of scenarios that people are testing to kind of trigger bugs and then yeah that's kind of the - two arguments in favor of keeping Rob stand and not starting a new proof of work test net was there any any other

things Peter or you saying mm-hmm so what I wanted to add is that at least coming from the rinkeby perspective after we launched rinkeby people started to play around but I would say it probably took at least half a year until it rinkeby actually saw use I mean real-world day-to-day use so if you launch a new test that before Christmas I don't really think it will have value because nobody will use it so there's a huge developer inertia because everything is deployed on Rob's ton all their setup all their workflows is set up for ups done and it takes a lot of time to switch over it so I think it would be valuable to switch it to start

a new test intent to just kill Rob stone but that won't really help us test Constantinople in my opinion so that can be used for the next heart for it any other comments on that yeah I agree with Peter and also so there's this notion in the comment that there should be at least one successful fork and a shared Testament I don't understand what is so particular about a successful fork in real time I don't see any value in that actually and if anyone has another

opinion please tell me what is the value of that seems like it broke really quickly after the fork and so just from because of that wouldn't it makes sense to have another fork or is that more something that always goes pretty smoothly the transition it's just if there's a bug then that's the problem so make sure if there's any bugs that we find them because imagine if this happened without a test net we got this breakage real light network I think we need these dress rehearsals but I mean anak find that successfully did it successfully managed to pass the fork right so so we have you sure you're saying we don't

need to test the fork transition part we just need to do a lot more tests with the fuzzer and other blockchain transition just doing a full sync to the fork I don't see the value of doing it in real time you know I don't see the difference between doing it in real time versus doing it on the historical data on the fork that order to happen any client yeah I mean there is a difference so it's Alexi here so the the difference is that there are different parts of code at-least is the code that I've looked at there are been activated

during the actual real time transition and when you just simply sink through the through the fork and one of the things for example is the when you do it life there's always these uncles for example which are coming from the old like a pre fork thing and I'm not sure how much of the problem that is but also from the other point of view is that if there was so much noise during the fork on the Robster we might have not noticed some minor issues that happen but because we were so overwhelmed with all the big things that happening maybe nobody notice that and if things were quieter we might have noticed them so I think I I mean at least I would feel

more comfortable if there was kind of successful work for what happened is not very successful well yeah but essentially eruption is already forked and you won't be able to spin up a new test not with him with decent traffic well yeah that's why I suggested in the beginning to roll it back but now it's probably too late so but we can still do it on a different one exactly so so if we keep Rob stand for testing the actual transaction and stop happening on a chain and if there is a group who actually wants to do a live transition then we could just get up I mean we can

spit up a new test net like that's correct yeah that that's what that's what we should be doing I mean until maybe with just maybe with no usage but actually just getting through that transition and then if we do not notice any issues then it's good good result yeah I mean that doesn't even have to be a large coordinated public effort really it just means that yeah we need a Terri MJ turbo gas I guess yeah parity and whoever else wants to join in and we decide yeah kick it off on like Monday or next Monday that it's been for an epoch or two and then do the

transition and forget yeah yeah that's that's it that's correct yeah that's great yeah I'm on board with that and this would be on Rob stone no no no it's just basically it is especially launched Network specifically for the transition test and we can just repeat it if something doesn't go well they can just repeat it again like one or two weeks later and then repeat until we actually get a smooth transition got it got it okay that makes sense cool yeah that sounds good anybody have comments on that yeah one for the question on this and so will this be a proof of where

mini chestnut and and would it make sense to then use this in the first weeks for Constantinople testing in some semi public space and then enlarge this to a new to a new proof of work tested or to have these two thing to combine or what wouldn't make sense really no they're nice just to just throw it's kind of throwaway chain to be honest so okay thanks I mean if it's so it must be a proof-of-work chain because that's the only thing every client implements but if we start with a proof of work chain and play around with it we might as well

keep it eventually so there's we can't keep it or we can throw it away I don't care today yeah that's true okay I think that's a good idea can we have a group of people who are kind of lead on that as far as getting that done in the next well let's say I guess let's figure out the rest of the timeline and then we'll figure out the timeline for for this new testament machine awesome thanks Martin

as far as timing for the rest of the stuff as far as so basically offer he just kind of threw a date out there middle of January January 6 tene ogre and the event that we were to start a whole new test net which sounds like we're only doing from a you know limited testing perspective and not a widely used test net he was saying fork on the twelfth let's just throw out some dates here so I guess the facts are we've forked on Rob's ten we had some issues but ever since those issues have been resolved we haven't had any major issues on Rob Stan with regards to the fork and we are starting up a new

proof-of-work test net to test some of the transition that's Martin Martin's leading and that's gonna happen before the end of the year is the assumption so what about what about dates what about the middle of January is that too much time is that something where that'll kind of be the holidays are over and so there'll be a lot more people on here what's the what's the thoughts here I'm sorry hey yeah this may be a good time to just jump in super quickly and discuss the difficulty BOM do we have 60 seconds to go yeah let's do that I didn't see any that I know that's in

here but I didn't see I didn't see that oh that is on the agenda okay yeah go ahead Lane what's up yeah Bom cool yeah sorry I added up the agenda I just it's super quick so I have spent some time working on this over the past couple weeks with a few other people J rush James Hancock ed Mazurek and QCD trio so we've had quite a few eyes on it we also discussed it with italic ran the numbers did some simulations and I'm pretty confident that the following is going to happen the difficulty bomb we're gonna start seeing block times increase in January early cuz probably mid January is when they'll start ticking up into the kind of you know 15 plus second range and then we're predicting that we'll hit

thirty second block times which is when the byzantium hard fork happened last year in May early to mid May and of course this depends upon the total network hash power so if that hash power were to drop by something like 20-25 percent and it could happen in as early as kind of early April so just put things in perspective we have time there's no critical concern but the main point I wanted to make is that it's gonna happen faster than last time for a couple of reasons one is that the overall network hash power is not rising like it was in 2017 it was rising very rapidly while the bomb was sort of ticking so to speak and also because the total hash power now is so much higher the total difficulty rather is so much higher when the bomb does hit it will

come faster so last year it was something like seven or eight months until we reached 32nd block times this year it's gonna be more like four months so just just putting that in people's minds that's it okay great that gives us a little bit of leeway if things go really wrong that's good and then alexei you had a comment on the timing no no I had just something that I think we forgot which I remember you mentioned that during DEFCON is that the specification the yellow paper yellow paper I saw that there the yellow paper has been has does contain some of the elements of Constantinople but I don't I'm not sure if it's been updated for like create you extra dust things so I

think we we need to kind of get it up to speed up to date as well yeah like officially adopted yellow paper and anyway right yeah we haven't officially done it there's just been some side conversations at Def Con about it but we haven't officially adopted it as the main as the main thing anymore but the argument there is since EOF t' the foundation and is no longer involved in aetherium that we know of he's no longer keeping up with the yellow paper and the jello

paper is more up-to-date and deals with K which I don't understand all the technical stuff behind it but um basically it's it's like you can I guess generate tests off it eventually if you want to do that and you can it's just a lot more maybe formal I think is what the word Everett used so basically we're kind of saying that the jello paper might be a good replacement yeah just to add some comments to that so the jello paper when I talk to Everett the DEF CON it is actually more I mean it formalizes the evm execution part there are lots of

parts in the yellow paper which is still very relevant like block validation had a validation state transition and those are kind of big parts that need to be additive to the jello paper what's the reasoning behind creating something new versus updating the existing yellow paper to include those things yeah I'm not expert at okay if someone else wants to take a stab at explaining it and I think the baseline was that so the the the problem that people the

community have been telling us over and over and over again for the past four years maybe you know more is that the yellow paper is extremely dense and it's extremely hard to follow and the problem is that unless you are unless you have fairly essentially only a handful of people have the experience and knowledge to make any modification to that so it's I guess it kind of boils down to how many people in the world can properly update it so yeah but there's also another cool thing which is that this is written in K which is language and from this so the specification is also an implementation and you can actually take

this specification and subjected to tests and execute the specification and see the results of certain transactions or state transitions or whatever which is really powerful thing and then you can do formal verifications of things from the specification well yeah I mean I agree with all that it just seems confusing to have two separates efforts in different era like why not have one thing that has both of them but it's like I don't have enough background on this I'll just I was just wondering if anyone had like oh it was clearly

because of this reason but yeah yeah so I guess the idea is not to have two separate things the idea is to completely replace the yellow paper with a descendant I guess that's the long-term goal additionally the jello paper what the goal of it would be to add things that the yellow paper doesn't have and take out things that the yellow paper had but are no longer applicable because they change so quickly or so often so it would basically be a whole new so the it would have the goal of the yellow paper which is to have a common specification that clients can follow but have it be more modern and have it be up to date

with the current standards would be the overarching goal Fred all right so I was following one client team that is your own client from scratch and what they did so when this is a down side of the yellow papers they had to basically compile different versions of paper to implement an assumed client because you have like a frontier version of the yellow paper then you have like a homestead version then a champion version now it concert in open so I think Pro for the yellow papers that's assuming readable but the corners like you have no versioning on it

yes and I actually I just had a quick glance at yellow paper like for last two days and I did see that it has elements for all these Forks like Kedah schedules and stuff it's all in there and so that might be a solution to that so it sounds like we're kind of getting to the point where we're fine with replacing the yellow paper I haven't heard anyone in favor of keeping it honestly is there anyone in here who is in favor of keeping the yellow paper as far as the common spec well I would say let's see where the jello paper gets us and let's see let's wait until it's action usable state and decide

afterwards yeah that's a good idea that's a really good idea to have both like on one hand you do want a mathematical spec that someone can look at and verify and review and on the other hand you want something for implementers to be able to quickly create code from and one is not necessarily suitable for the other so I think there's there's value in having two I think the jello paper fits checks both of those boxes but I wouldn't stake my reputation on it and I also would wanted to say that I think that they I don't know what like what is the who's

actually working on jello paper but I assume in at the moment there's nobody so it has been but it kind of suggested as the specification but I don't know do we know who is actually working on it because we can't simply wait and courage to to fix itself right are you talking about the yellow paper or the jello paper jello paper I taught me know that nobody oh go ahead I know that nobody's working on yellow paper but is anybody actually progressing with a gel okay yes Evert told me that they're gonna have someone dedicated from run time verification to update it and that they're gonna start joining the all core death calls I think

they're off this week but um yes and additionally they they plan to actually build well there is action implementation of ethereum client which uses the kay engine and put that through the high tests and put it through the Lib foster stuff and actually physically vacation so their plans pretend that they're maintaining it okay very cool great so we don't need to decide on this today at all but we'll see what happens when the jello papers updated will start getting runtime verification and the

calls and go from there so back to the hard fork timing do people have opinions on that otherwise I can just say what I'm thinking who wants to throw out dates or I guess yeah dates so this is like four days before the East Denver conference it might be better if we launched it a couple of weeks before in case things go poorly which date are you talking about February yes assuming we don't find another full disclosure the last consensus issue was

found 16 days ago all right quick check here so there's been some proposal to to move things out but our original plan was working mid-january if anybody in the call who is who's the advocate for removing things out past January and I suggested of setting up a new new New Testament but we we went through this and I think with this new plan of having just a small from dedicated test net that it makes very much sense to keep

the January a date so I would drop this a February suggestion well does anybody else want to speak up for moving it otherwise it seems like our default action here should be to retain or existing for like our previous plan of forking in mid-january mid-january seems fine it seems fine to me if we get tests ready like a month before that need according to the me tree I think we're on tap on track to get it ready at least a month before and there I think most of them are done is what he said he

had he said he only had a few weeks to go to finish the rest of them when we had our last call about a month ago so I don't know if I was able to but yeah yeah yeah I can say something about it because I just have taken on some some additional responsibilities they also work on the on the chests repository and doing some additional coordination there and then help Demetri bit with with all the other stuff none implementation-specific um yeah and actually we today Dmitri merge the Xcode of initial Scottish tests and other tests are now in the work from Adrian from Pantheon has joined for helping out and to go from Eve awesome and I think we're we're on a very well

track to get these tests done we also just one other thing we we changed is that we now introduced a releases for tests for the tests so we started with it started with a version 600 beta one and so so that the client can actually synchronize their the testing efforts and we can compare better the different the different test results and I think we should go to a kind of like main release or like six level release and in the next two weeks or so yeah I think there's this one a good trick so we got to testing one thing I would really like to highlight is that the test Suites are

not so you cannot really rely that they managed to cover older really with corner cases there have been ample cases where we reorganized some virtual machine stuff within go cerium and all consensus tests were passing and then syncing may not just failed so the tests are it's not a bulletproof thing that if you pass the test then everything is golden and from this perspective um I think one of the most important features that we need to support is actually the fuzzers because if if you your client is compatible with the fuzzers then we can run it among straight against other

clients and see what it finds and as far as I know most consensus issues were found by the buzzers nowadays yeah should I give it to an a second walkthrough what's needed oh you're muted Martin I was just wondering if anyone wants front of a short description of what's needed yes yes yes yes please yes so in order to get first there needs to be a a an executable for your client which takes a state test as inputs and outputs for every opcode a JSON L a JSON

line object the details the operation that's happening in a format which we call a standard JSON output it's documented on a VM lab it contains things like Oh code a number of name astraying gasps certainly a couple of Morpheus the stack contents and stuff like that and it ends it all sneeze out but state route after executing the state test and that's basically it and it's preferable if the clients spit out JSON line of JSON line line by line

because if they do it in one big JSON chunk afterwards it will just blow up the poor Python framework that tries to run a lot of tests in parallel and which alliance go ahead Martin now so currently it's gues imperative I know that Alice has kind of those capabilities at least to spit out the common data format but as far as I know it cannot execute on a rural state test in order to do that we can run it in testes but test if as far as I know spits out and a big blob which doesn't

work well if there's 20 50 thousand okay that sounds good many questions on that from anybody or comments okay so back to where is this ocean online where is this online where can we get it so the framework is EBM lab ethereum CVM left and the actual production Foster's are running on the server Martin you said that there was documentation for it can you maybe drop a link into the Alcor devs for the documentation for that

format yep sure okay great so it sounds like there's gonna be a little more coordination on setting that up and I and I think we all heard Peters point that there is more value to getting the fuzzers up and running across clients than getting the manual test going even though the manual tests are also important and they're nearly completely no I don't know maybe I didn't say it's more important but I was saying that so passing a manual test is not yet out of the woods so you still the fuzzers can catch a whole lot of things that manual tests cannot or will not got it edge cases okay so it sounds

like we're kind of mostly agreeing on mid-january for the for the date do we want to wait till next meeting to decide an exact date when we're kind of seeing if there's any more issues that pop up or do we want to decide it now I guess what's two weeks from now that's like mid mid-november or so that's Thanksgiving in the US big weekend holiday is that Thanksgiving Day it's actually the day after which is Black Friday which is a shopping holiday that's true shopping holiday wow yeah I mean let's not be too us-centric unless anyone has strong objections I don't

think we need to reschedule that all right I think we'll be fine cuz most of our teams in Australia okay yeah I'll be fine I think I don't know if I'll have another monitor to actually do the recording maybe Lane can do it I should be able to okay wonderful what will coordinate on that okay so it sounds like mid-january is what's going on I propose that we wait till the next meeting to decide an exact date based on running some more tests and seeing if anything pops up so that we would need to delay it if anything pops up that's major anyone think we need to decide on the date today or is there like a

benefit to deciding today I guess we can just say B January so even I mean it doesn't really make any Frantz if we decide on a date or not because if something major comes up we can always postpone yeah I would say that let's set a date for today and we can always kick things out next meeting if something changes right now it seems like we're in a spot to be able to set a date great how does January 16th sound that was the one that was suggested I mean that's a Wednesday which kind of is nice because if mining power goes up or down a bit

it's still far away from we can so find by me can we does it make sense to move that to like the 15th which is Tuesday which gives us a little bit more of a buffer for mining power or wait the mining power goes that makes things earlier doesn't it nevermind Wednesday sounds great great anybody opposed or have any other ideas I like it okay xvi it is yay to be confirmed on black friday to be confirmed on black friday correct great you have sent your

Christmas gift yeah looks like we'll get it for 50% off to you the only thing that I think we should also highlight that if we're going with the 16th then ideally we should all clients should release stable version with the baked in block number before Christmas I agree yeah okay we can all make sure we do that I think we can get off this topic unless anyone has any other comments okay good job everyone

we came to a decision the next thing is item five Prague pal update who wants to cover that one Powell maybe well from my perspective we have three deaths from party from from from Goa turbulent and myself that we're working on implementing the CPU part of the program CPU implementation let's call it this way and and we kind of have agreement on what what's the output the results are like we are in the sink but the spec is

not finalized yet and there were some changes recently to that and some some tweaks and tyrannous and so on so we are not we are not directly wherever the spec is so this is like one of the issues is spec is not finalized not also because properly I described at this moment and it doesn't include any test vectors so mmm that'sthat's main struggle we have I think but besides that I think we are in good condition on this on this yeah in this in this part some other related infrastructure updates would be we just

proposed this is by Andrea from if minor community she mostly has written respect for a cheery on Stratton protocol that would include data that is needed to properly do a hard work and to properly support protocol in general yeah that's the second front and I think that's mostly after her okay anybody else have commentary I just wanted to say that it

was a actually talk at DEFCON four I think it was an ultraviolet I just got walked injured accidentally that was a presentation about profile and I remember I sat through it and they included the saturation numbers I think the ones that we so I was suggesting they would produce so they actually had saturated numbers in the presentation and I think they were pretty high so yes but I think also that the presented during the workshop is also in a public in a medium post that was published a week or two before earlier so I think the numbers can be found publicly there

if anyone has the link you can put it in all Chordettes for people to see yeah the talk was an ultraviolet Oh got a girl presented and those videos for dev cards should hopefully be out in the next two weeks depending on if my internet works again I'm about to give up on my ISP and move to a different one but the issue was like that the other ISP also sucks so like my choices are very limited so if I can get my ISPs all figured out and get my upload speed above one megabyte per second then I can upload the videos so that'll be good so how does any if I'm understanding

correctly proc roof of work is delayed because of your internet connection that is exactly right to do that article title what's in it would be faster feel free to like FedEx me a pin drive and I can upload it from here okay okay yeah I might I might get you or someone else to help that'd be great Public Utility hack is to try going to your library not bloating from there sometimes they have a giant pipe okay you're a university universities are often open and very fast okay yeah off to do that alright so it looks like we've gone through Prague POW and gotten an update on that

the last item was the Gourley test net which we've already gone over and I think as their other comments on Prague pal with their other comments on the Gourley test net or anything else so I proposed two things to discuss I'm not sure what do you want to get to that when you just get them I just didn't see them go ahead and discuss them okay so one of them was that our API is the RPC api's eath underscore network currently returns three parameters for ET hash my name and we've been approached by a few mining pools that this is problematic because it's a both parity and gas when

they start mining on the block periodically they recreate the block even though nothing was mined yet this might happen because a new uncle arrives or it might also happen if periodically you just receive a ton more transactions and if you recreate it it might actually be more optimal for you I mean for the minor and for that for the minors it generally helps to know whether this is a new round of work and they should quickly abort mining on the old one or whether it is still the same round just an alternative block and the solution that I think parity what has been shipping for a long time now is to add a fourth parameter basically just a block number and it seems to work out

fine ish I mean I can't imagine a few corner cases but generally it's fine and it's an easy enough solution and we also wanted to add that to gas so of course we can add it and we added it but since there are actually two more already two clients on the network which work according to this and it's generally useful for everybody I thought it would be might as well just update the spec or json-rpc spec so that it actually requires returning a fourth parameter when you say expect to you me in the in progress EEP or the wiki spec now the weekend so the wiki RPC thing is just adding two

more lines to the wiki page the reason I wanted to bring it up because it's at standards documented API spec so if I if you want to change it to be nice to have a agreement on it no objection from me I think it would be good to to make sure that that change does get propagated to the to the in-flight EEP for speculation RPC AP ice okay I'm not really sure what which that is or what what the processor is on but if you can ping me deep from where I can just write all I'll link it in chat and and yet I'm literally talking about but

make sure that there's a comment on there to indicate that it needs to get updated that's all okay I can do that if you like it so anyway that was one of the easy issues and the other one is that we've been having discussions on at Def Con about setting up a few cheap processes within etherium essentially for example we have absolutely no process on how we are doing Forks so it's kind of these ad hoc things where everybody just we just talk on chat and try to do something meaningful and then hope nothing blows up which didn't really happen during Roxton and then the idea was that it would actually be nice

to try to somehow define a few processes around important repeatable actions for example the glowing an artwork and we've had we discussed it a at Def Con and I've written up a small document I also faced a different Accord of issue github issue so it's a proposal that I would like to it would be nice to have this playbook so to say that okay if we're doing a hard fork how do we go about it and in in my my suggestion it's a proof of concept again I kind of highlighted that for example it would be nice to first do a do a simple hard fork on a on a private network so to say that client

setup between each other because then we can tell these transitions and catch obvious errors blog def dependent errors without messing with everybody and even afterwards if we if we decide if you're going forward of forking a test that Lane had this brilliant idea at Def Con that instead of doing a live for covetousness we can do a shadow fork whereby we set up a few forking clients that just fork away whilst still leaving the real that's not on its original path and this allows testing quite a few nifty things for example testing whether clients can cleanly separate from the other network or if if there's a stronger network

going then all hell doesn't break loose and there are lots of these scenarios that I described in this document so my request is just maybe have a read at this document figure out whether different or other clients or ecosystem participants would be on board in doing such a document and trying to for example have hard fork follow display plating playbook and if there is a traction then we can we can possibly work out all the kinks and make it expand it to whatever ether we need and one more thing that people were quite a few people asked me why why there's a

need to do a separate process thing instead of just filing an EIP mm-hmm and I don't this is again debatable let's get the document on first and then we can see whether it's actually fits into any yeah IP or not my initial idea or initial reasoning is that the IP is more like a specification I mean kind of functionality spec but it doesn't really tell you how to do it whereas this process documents the goal would be to have a step-by-step checklist where you can just go through and just tick the boxes that hey did go aetherium I don't know report to the status page no then we're waiting on tilde reports then

these kinds of boring stuff that aren't really specification rather just mundane tasks caters if we try if we want to try to use this kind of workflow we're then it'd be good to have some kind of appointed director whatever Forks on the point so that's actually part of part of the consensus and part of should be part this whole process document that if we'd if we decide that the hard work needs to have a heartful coordinator and in my

opinion it should have one then probably we can also add that to the to the doc that yes first step step zero is the point somebody who's willing to coordinate the entire thing I totally agree with that the only ideas that I think it's a if we kind of agree in advance to some document on how hard Forks are going it's everything can go much smoother because everyone's on the same page and everyone knows that okay then in two weeks we need to launch a private network and in four weeks we need to do a shadow for cardfight of the rock stone and then you you have a clear timeline and you have clear steps whereas currently it was just hey okay let's that select this block number and

then eventually hopefully everybody will do a release and hope for the best so that the parent didn't work out we discussed a lot of these ideas at the I guess the East 2.0 gathering it was called in Prague and there's definitely a lot of support there seem to be consensus that the coordinator role makes a lot of sense so I just wanted to add my voice and or I guess the voice of all the people who were part of that conversation of course you know figuring out who that person should be in the process by which they're chosen and whether it's one person or a group etc is complicated and I don't think we should take up too much more time in this call I think it's a great idea and Peter I think that you know checklists if you look at other

professional engineering or you know fields like medicine or I guess that the canonical example here is like the pilots checklist in aviation I think it's a fantastic idea and I love the idea of having a meta conversation not around this particular hard fork but around sort of a platonic hard fork right what that should look like what the steps should be the roles etc and one final thought which is you know we've discussed many times now splitting already the AI sorry the ER C's out of the EIP repository which I personally think is a good idea I personally think the EE P should be separate as well one possible way for us to fill this role would be for companies in the space who have existing knowledgeable project

managers on staff to donate one of them at the time that we are entering into Hartford range so if there's companies out there who would think that they've got somebody who would be good at this and they would be willing to donate to the public good that might be a good way to do that so I I don't really understand this line of thought with the project manager thing and somebody being good at it because I guess the whole point would be to create a document where everybody can be as dumb as it gets and still be able to say tell whether you're following through the thing or not so that's the whole point that you don't want smart people who make decisions you want a document that it just tells you what to do and if you

follow it to the letter then it will just work out fine sorry I'm not suggesting somebody who's a decision-maker exactly this is coordination so somebody who's keeping track of whether we've done the things making sure that it's clear what comes next all of those things but that's it just helping all of us who are busy doing things that sometimes we don't pay attention to the process it's nice to have somebody to help hold a hand every once in a while I publicly asked if anyone had that set of experience and wanted to contribute at one of the 'if magicians events in Prague and I had four or five people I think reach out subsequently and and say they'd be interested in helping a number of them

said they'd be interested in volunteering their time Hudson and maybe Piper it might be good for us to set up a call with some of those people or at least discuss amongst ourselves offline and see if we wouldn't want to accept some of that help yeah I think that's a great idea I'd love to do that cool I'll reach out offline about that anyone else who wants to be part of that conversation just let me know yeah and so so what does the acronym again Peter is an EE a yeah I just called it so that I call the repository to police just um engineering effort and documents I just

call them in theory imagine you processes okay Tripoli names but you can I'd found to believe to be a fancy name I liked it mature I liked Tripoli too personally okay so we kind of discussed a few different things at once here let's go over the first one real quick we have about 15 minutes left or so so updating the F get work RPC API on the wiki did everyone kind of agree on that I kind of missed that part of the discussion yeah I think the main issue here is that in the past that wiki has been completely ad-hoc edited by anyone I think it's still publicly editable by anyone so I

think it's more it's following sort of due process to notify a I think this should be changed on the wiki and then change it once there's agreement so I think it's more we're now starting to do that it doesn't really fit in within the AIP framework but I still think it's something we should do and I think yours quite doing it right away but I think that there's not it's not a big issue like it's not a big change or anything so and from Perry's point of view we already implement it so from my point of view it's this is very useful information probably so it affects how

we're going to do a hard work if any to two different proof of work algorithm and also propo takes the block number as their as an input to the hash function so this is almost requirement for that and also block number is much more useful than the it's called C - it's pretty damned private there so even more more aggressive approach that would be replaced a sit hash with block number but that's probably can cause more bugs for backward compatibility issues so I

fully support it great it sounds like we're all good then and then I think everyone was kind of giving positive vibes behind the eat tripoli so let's go ahead and try that I don't see why not a worst-case we abandon it if it's not working and then we'll get with Blane and Piper and others about the project manager position I know there were some other people who approached me via email and in person at Def Con so I have some names as well and then was that what did I miss anything

Peter okay any other stuff Lane you mentioned potentially separating the ERC's from the e IPS did you want to kind of go over that a little bit or is that for a different day yeah this was arc about that on this call we only have like 10 minutes left and that's a big thing at least that's my opinion do I'll give a quick summary lane then yeah let's just do that yeah I'm not super keen to go into great depth on this I don't know if this is the right moment or the right forum to talk about this but I had there was a conversation about this I can't remember exactly with whom or exactly we're at DEFCON but I mean the idea is like

separating consensus critics like stuff that falls within consensus from stuff that falls outside of consensus it's not at all clear to me that those should be the same process that the same people should be involved etc and and I think that that they all live in the same github repository for historical reasons but I would be for separating the two of them cool so we'll have that on in a future call potentially or if it's not the right form maybe magicians and we can decide from there with the EIP editors what we want to do that sounds good any other topics or comments or anything like that awesome we will meet again in two weeks on Black Friday everyone gets

some good deals out there if you're in the u.s. if you're not then you're missing out on a wonderful United States made holiday for consumers all right I'll talk to everyone later everyone have a great day or night bye thanks Hudson thanks bye Singles day on the 11th [Music] [Music]

[Music] [Music]