Status Core Developer, 6 Aug 2018

Previous Meeting

⏯

youtube image

►

From YouTube: Status Core Dev Meeting August 6th, 2018

Description

This is the first core developer weekly meeting held. Agenda is posted ahead of time and details on GitHub.

https://github.com/status-im/pm/issues/1
Notes (WIP): https://hackmd.io/8uQR4O65QNWwjSx3iPN9Rw

A

Thanks so much everyone for for joining sorry about all the technical difficulties here.

A

So this briefly sort of the purpose of this meeting is that we have a lot of people at status and there's a lot of meetings and so on, and the idea is that we're doing lots of cool stuff, but there's some sort of cross team and so on collaboration I, would be useful, so didn't idea that everyone here can sort of talk about things that we're working on and we can have a bit of a discussion and see where there's syncing up points.

A

Maybe some research that's relevant for the people or maybe there's some client updates that impact certain themes or efforts, and so on and additional sort of use of this is that people can consume it after effect, so that we have like this recording and we have notes and so on. To top you can refer to and initially hopefully we'll be able to invite other people in the community and experts in the field and so on. Who so because of talk?

A

Was specific integration points and so on, and it's kind of loosely modeled off today, theorem all their meetings that are every other week and yeah roughly like that, so it is sort of a technically focused and not sort of presentation like Town Hall, but sort of more of a discussion, and we can, even though there might be someone who talks to it more about some specific point. That is that we can have so general discussion about various topics and so on.

A

um So just briefly, first, let's talk about infrastructure updates, and that includes general infra as way to be a layer, security, interesting and then client updates, mobile and desktop. Then maybe after some recent updates and product area updates and then specific ideas and miscellaneous stuff, that's roughly this format and then at the end, since this is a very new we just wanted to check with everyone see, people for this was useful, not and any feedback people have probably can improve it before we get started with instructure uh places.

A

People have something I want to add e to agenda or questions or anything.

A

Right all so I guess Jacob. Do you wanna get start off with general infra optics yeah.

B

Sure, okay, yes main stuff, that's been happening as with Jenkins I guess effects everyone. The master has been placed and I'm almost ready to replace the distilling. Linux slaves I was just working on Friday to get the gist of those work properly in Linux and remaking their their Jenkins file. So that should be done today and the old all slaves with far too small disk space will be gone and also have a proper monitoring for all Jenkins now she's, except that McCoy's was because those are tricky.

B

I'll have to think about those, although with help from Romagna, and that we have one more slave for building the first one still doesn't work, but we might get that done. We'll see the main problem is signing with Xcode and that's a essentially black magic working something's. Don't in other news, I, don't know about the release happening this week, but I have the ability to other Third Fleet in Hong Kong for Alibaba cloud and I.

B

Wonder if we want I, don't know if we want it for this really is this week, but maybe next week, so we can start testing it. We can discuss it later in infra core, but I'd be happy to just add another fleet for just Hong Kong. So we can have all the three major continents covered like area in Europe, United, States and Asia. So.

B

Anything else that is concerned most people I, don't think so. That's the majority of it. A lot of cleaning in in logging here for cabana result of noise from various services. So I've dated a lot of filtering services and I've changed the formatting for how we said locks which is not JSON, I can add any field you want in an aerosol configuration, so logs should be clean. There now in Indiana feel free to just dashboard sad, whatever graphs or views you you might need. Its power is.

C

No, since you asked you know this week, we won't be able to add any fleet for the reasons. I will talk about the client update, okay,.

B

Thank you, that is great, you know, cheers oh.

D

And I have question about the staging fleet hasn't been tested or because I was like briefly looking into that and there's a comment but I haven't really like. Let know other people that it's ready so was it tested by by the by the testing or or not really.

E

For sure it wasn't tested but I test him.

D

Okay, cool yeah, so I guess we should probably oh because there's a there's a still an issue in status. React that's about selecting a fleet from the UI. Yes, exactly I! Don't.

B

Think we have an easy way of selecting a fridge right now for testing yeah.

D

That's true: okay, I got it yeah cold, so yeah I've known only testing this, which stuff is go but yeah cool. Whenever this is ready. We probably should also test it like, probably because, but.

C

Simple thing: just raising one more job on champions that will build nicely with just stay in sleep and I mean we can do it again in a few hours right instead of making a UI that will take a few days and then we'll do the UI actually.

D

Yeah we can, we can think about it. I mean because, like from the from the Batman perspective, looking at least from status go perspective, I tested, it was fine I'm, but yeah. We do not have any nightly or like any political tests. That does that so yeah. Maybe that loves a good and that's a good call to actually create.

B

Food selection doesn't have to have a UI element. It could just be on a file yet somewhere on the file system that selects, which fleet to use when using settles go. It doesn't have to be in the UI yeah.

D

It's maybe yeah.

B

D

We'll see what's the best Brad.

A

Cool right, yes, it's anything else on info. Should we move on to security, I.

B

Think I covered all the major topics, there's a lot of smaller things, but I love the instant eye for that all.

A

Right so a little security, so when a people haven't had time to document with interest Corey who just started working as a senior security security software in there this week, Corey you honest, guy, yeah,.

F

Sure, hello, everyone if I haven't, talked to you I'm Corey gonna, try and take over all security initiatives, at least like lead them across status, which includes picking up the champion program and making that more robust, so that the overall culture of status is at least asking the right types of questions and, generally speaking, I can't fix all the problems. But I can try and help give tools and provide tools to people across the building teams that they can make better decisions because they're the domain domain experts in whatever they're doing so.

F

As long as we continue talking about these things, I'll try and provide tools that help people make better decisions about security from an overall perspective, whether that be securing the infrastructure, making sure that people who are trying to get at them can't thinking about changes being made and whether or not they break something or add vulnerabilities or cupful miller.

F

Fix vulnerabilities like just maintaining your own, your own personal security, better, as well as making sure the end user has an intuitive experience about what the differences and status are compared to other messaging platforms and how to like. Keep their digital valuables safe and secure using status least that's the goal.

F

You have any in particular questions for me or anything like that. I've talked with a good portion of people so far on how we're planning on doing that. I'll have a kind of a introductory blog post thing that I'll publish here pretty soon. That's gonna be kind of my overall viewpoint on these things and what I think is important and how to go about doing these types of stuff. But until then it's it's kind of like that. Hello.

G

We have different, we have the desktop stay. The mobile mobile, so I think they disappear in here. It's about the mobile version and what would be the case is I mean the user need to be rotated to have some issues or what will happen? We have some idea on.

F

On the mobile app we that mobile app is I guess each each thing has its own way.

G

F

A mobile it's gonna be along. You know, reproducible builds so that people who build or get an app understand that it's it's built correctly deterministic bills is going to be I. Think a big push for that to make sure that, like you, can check to make sure whatever you're getting is what you're supposed to be getting and then I mean III. Remember the process of trying to absorb how all the stuff currently works and how all the pieces of status fit together.

H

F

It's hard for me to have hard opinions on what needs to be done and who goes here to do it.

G

Modified version like a fake version of status mythology users, and for that use we can maybe have some kind of fingerprinting or side of that. That would help so movies, but I also wanted to know more about route keeping and right devices, because I see that there is some potential risk, keeping devices special and right.

F

But what's the what's, the question is that a problem or yes.

G

Yes, we do, we should look what what are they in the risk of repeated a droids being using state because they to saves the heat of the user is encrypted and by actual somewhere in the data of the app. So we will keep in devices. This could be no, not so fake, wait so which we should see.

G

Maybe it is to check if the app is in solid device that is repeated, and maybe even while the users about this I see that there are so two device from sorry applications that doesn't allow should be running little devices. Maybe we should only wire them.

A

Absolutely maybe we can, you know every encounter mode more like the specific fat models like that. Late rest, we gets with more hug done or granular sort of views on security and so on. I'm.

F

Hoping to set up like different different, like Brett hunts, vulnerability of sessions, things like that with each team across that, so that we can dig into like specifics of each a problem set that you may have is what what you're looking at and how you you handle security on a mobile app is very different than you know. Looking at the status go type of stuff things like that, I don't as of right now, I I can't I I want.

I

To look in the root page but I.

F

Think, having all of us together trying to ask the questions at the same time is better than doing it right now,.

D

Yeah I just wanted to add that that was actually tested during the first audit. So there I saw that company. They actually run those tests on the route like yeah, like locked, whatever Android version and our app actually signals this when it's run for the first time but I, don't remember it exactly, it might be a problem when you install status there, you unlock, Android or allow you to get in whatever and then run the status again. So then, and when the only running again it just not notify you about it. So there's like it's.

D

There might be an issue of that, but but overall we do notify users when status is run for the first time on a unlocked Android. So just just a comment about it.

G

That's all about security, but in status we also going to have this multiple, like a relay on the on the security of the that's really quite fun. Because of this we also developing the hardware wallet, so the user should have like this male.

G

Like this would be very unlikely, sir Sir is how these layers of security one is. They have target application with the cell phone and the other is Hardware wallet that works together. Oh if you need to be like a fake, hard wallet and uptake application to make it bad bad action instead of adoulin security.

G

For for us to have this, we don't redundancy in security and also think about security with my counter to enhancer but I. For now, I connect status is going forward, a good blue, a good security model, all right, Forex users.

A

Would you guys should we move on because we're quite a lot of things to go through and I think we can talk about this this a lot, but maybe it's useful if we have like, if it's more of a specific framework, and we can talk about things that are more useful to lots of people, what do people think what that.

D

Yeah I think those are body concerns, but yeah I would move them to a separate meeting. I'm.

A

Sorry, good yeah.

E

I also agree that it's a lot of things to discuss like especially with priorities like there are some parts of products that being devolved but not yet been used by users and some that already released and is using by why the audience so setting priorities and making a separate sections would be great. Because yeah sure we will not talk about it and mainly 30 minutes or so, and.

F

I'll be doing a lot of that like very soon and trying to set up those meetings. So we can have those conversations in death.

A

Perfect I guess Igor. Do you wanna do a brief update on security champions, we're wet yeah.

C

Yeah, we did finish some fella or we will be pleased tomorrow. The training session for NJ and Gorham I. Guess tomorrow, is the last kind of education-related day and then we'll start yeah. First of all, I'll start moving stuff towards the quarry, so he will be taking over this. So it makes sense.

C

So we don't have like to like the people doing the same job essentially and yeah, but we all start so looking into like Leo and thinking about attack like vectors on the real PRS and real features that will come to both Woollett and the app starting tomorrow, hopefully looks from the security.

E

Training, and is it possible to share with more people awake and testing, for example,.

C

Well, I mean I so far we decided to since it was like just to try sort of we decided to keep it like more private, isn't probably shooting this company but think it went pretty well at least I heard nice like feedback about this. So maybe we can just do something that it's more like open to people who are just interested. So I just see me to take this point just to have less overhead, since it was the first one, but yeah I probably need to sync up with or about this first I've done there.

C

We go matter of course like. If someone are this resident security, then of course the trains and stuff like this station will provide, because let's stay at the point of decentralization.

A

Cool and then, but initially we had planned a second audit, I guess Adam dude. What's the latest on on the on it right now you know.

D

Oh yeah yeah, so we pretty much from prepared a draft that I'm gonna send right after this meeting that we pretty much cancel. This I mean cancel this postpone it to sometime in the future. So I briefly described our situation and what we talked about in Basel send them a few links, so they maybe can also understand it better, but yeah and just it just from our perspective. It doesn't look like any doing it now, making sense.

D

It's gonna be lots of time for us and for them from their perspective, doesn't matter, but from our perspective, just like losing money and standing this depending them on something. That's not gonna be really useful. It's gonna change, so so yeah. We pretty much want to do them that we can do this in September, and- and here we are gonna, resolve the issues and then we'll figure out. What's next I.

A

Know I mean just briefly two points of context. Initially, the reason for for having that plan audit was because we wanted to push sort of a public release before Devcon and given the ocher process this quarter, it doesn't seem to be that much a project anymore and secondly, because we're changing around with the messaging protocol and we're gonna push up some kind of spec that we'll talk about later. It's probably more worthwhile to first have that spec reviewed on a theoretical level before sort of diving deeper into the specific implementation and so on.

A

Just some additional context. Sorry Rachel.

J

I think I think.

K

You cover there I just wanted to know here was going to be another target date to begin the security audit, but it sounds like it's just going to be based on one. We have those some.

D

Of these issues resolved yeah as far as I know, there's no specific date and what we want to postpone this, but yeah that's pretty much about making sure that we are actually ready. So I believe that maybe we'll we'll try to repeat the activity from from Basel and figure out if we actually ready with the trade offs that we have because there are, there are always beats on trade offs. So what are?

D

They are okay and we want to push for 1.0 or something like that or not, and also I believe that protocol itself, the the Masonic protocol is gonna, be audited as a separate entity. So I guess, like briefly, we talked about kind of to stay shouted, so he outed the protocol independently from them from the even from the implementation or the implementation itself and later how its integrated with the status. So there might be even like free rounds like theoretical one and two implementations like a implementation and integration.

D

I don't know yet, but the such ideas were were like I mentioned it before.

A

Cool so I guess moving into peer to peer and talk about the new peer discovery, protocol, Adam or Dmitri.

D

Yeah I mean okay, so so I wanted to like briefly talk not only about the new protocol, but also like a general status of the peer-to-peer network status, so just to briefly like just I'll, just walk briefly through the whole thing right now, so that everyone understands because I believe there's some confusion and how it looks like right now, so so yeah. Let me just look quickly briefly describe it. So so we have this status cluster, which is a bunch of servers running in free data centers as far as I know and three different providers.

D

So that's cool from the kind of censorship being censorship, persistence, I, even won IPS, or one data, centers blog just which happens in Russia. Then we should have at least one which is still working and yeah. We random, like good note, whisper notes, may servers this kind of stuff and they are all together like connected together. So even though there are like three data centers and three different providers, all of those should talk to each other, so it doesn't matter to watch.

D

One you're connected to user does not know that so yeah and later we we have like status network pretty much, but what we can call it or describe it, which is totally separate from a turian network. So we have a patch in like on Goa trium. That creates a pretty much a different version of discovery protocol and we also put a check and it checks if connecting appear, has the same version of discovery protocol and if it does not, it just disconnected.

D

So it's not possible right now to connect to our network from from outside, because that peer, which is just gonna, be rejected. And similarly we cannot connect to any nodes like to find any notes from the atrium network. So we pretty much have like a separate separate network for by our cluster and which is used to communicate with all people that are using status app and it's. It was done purely because of efficiency. So there's like relatively small number of nodes that you need to keep track of.

D

So it's much easier to find nodes good nodes, because we also maintain our nodes with high uptime and and the communication is much more able than it was before when we were using this whole internal network yeah, and that the drawback is, of course, that we break our principles. So the centralization, as you are, trying to do that on our level and make and use different regions different data center. It's still kind of centralization, because we run this and- and it's not like anyone- can run a separate note and start connecting to us.

D

They could do that, but it requires to not run and again and a few note, but our me yeah pretty much our version of a dream. Client.

D

um Yeah, so so so what what we have been doing, percent and and what Dimitri has been researching, is the new peers discovery protocol, which is called rendezvous and is based on the proposed leap, p2p specification. This is pro request to leave p2p spec repo, and it describes very simple but very efficient peers, disturber protocol and I've seen like a couple of other issues or requests that people are trying to implement this but yeah.

D

We did it as well and we wanted to use it in in the future in the future versions of of the mobile app. So we believe that that this protocol can be used as a kind of default protocol. It's gonna just just inform about the nodes that the part that the app is interested in, for example, I, need whisper nodes, I need my server nodes and run the vapour level can discover those and send them back to the to the mobile app in a very efficient and fast way.

D

But we also would like to to run our server nodes to support that new discovery protocol, but also support the atrium discovery protocol and and remove the patch that that were blocking external nodes to connect our nodes. So we pretty much so pretty much. Our server nodes would become part of the neural network, but they would also support this new peer discovery protocol. This way we would not sacrifice any efficiency on the mobile side.

D

That would be still super fast and reliable, but our our server nodes would just become like internal ethereal notes and they would operate normally like and they would not be really distinct distinctly from the other interview notes.

D

This is this is really important, with Alia support, because they're what you would need to have a year support easily as servers, so we could run alias servers now around, but that's another centralization level that we would need to maintain those and also scalability problem, because running Callias servers is much more difficult than running whisper nodes, so he probably in the first version of a year.

D

Support will probably want other people to just try to find alias servers from that room network and and therefore that we, for example, for that we need our our our servers of server nodes to be part of the whole single room network and not form like, is something something aside: something separate yeah. So that's that's pretty much the story. I.

D

Some comments on chat, so let me just quickly read them: we're excluding cows are somewhat removed that right. Yes, currently, we are excluded from from the theorem network. So we, our nose, are kind of forming a separate Network, even though they are using the same protocols. They are they're totally separate entity. They are not connected in a way to that room, a trio network and even said that this breaks his heart, yeah and I.

D

Believe that's that's one of them, the things that we did and it's really violates our principles and rules that we talked about recently, so yeah I think like we should get rid of that as soon as possible and use this new render for pure discovery protocol that Dmitri created, because it might resolve the same issue that we because yeah. This is the issues that we had before and which resulted in a in this separate cluster and yeah, and they provide this. This protocol can just provide the same.

D

The same goes the same properties that we right now have a separate cluster, but not not having this separate cluster so so yeah, but that we need to test this further but um but yeah once you want to fix this and there's a plan. So hopefully it's gonna be do.

A

It yet do we think this will also we do you think this will also be useful for other people, other teams in the space and as it's possible, that this could be an EIP.

D

Like this, like, then, the new peers discover protocol. Do you refer.

A

To yes or yeah, yeah I think.

D

A

Thing yeah yeah.

D

I believe so, like the whole description of the solution might be as something that that could be useful for people who want to run mobile notes pretty much or like all our notes that are restricted in terms of resources and and secure things, yeah I've. Definitely, but, as I said, the the discovery protocol is not something that we kind of design from scratch. This is implementation of a proposed leap, p2p spec, that is called.

L

One comment is a network implementation is kind of similar, but with little weaker proposals of small differences. Okay,.

D

So it's probably would be also cool to to maybe leak like, maybe describe them, because they are I guess they are like implementation details right. The whole concept of how rendezvous works is the same I just there like some implementation details right.

L

Is the same, but there are differences in ideas are different, so same link, civilization format, so we're kind of implementation, details I, see.

D

Okay, yeah, by the way.

M

Sorry yeah, but Sasori I mean the noisy place so I'm trying to speak as minimum as possible. So I've been reviewing a couple of talks and slides for some discovery of the v2 and defined summary, and they mentioned this explicitly. They they open them to experiment and maybe like try in the future. We p2p approaches so I think we would all benefit from using random.

M

Oh if it's like, if it doesn't have you know, Joel beds, and so it's not only for mobile right, so I think we should push for for including it as a deep world discovery mechanism for a Tyrian itself in in this way, we will not separate our our cluster from, but you know it involves probably write it some posts, persuading people explaining maybe given talks on why it's but etc, etc. So it's kind of more broader for than just implementing it. But I think this is where we should put a lot of effort.

M

I

You want to hear some why or what the Atlantians lament is introduced. You can generally also dance, but unless, unless it seemed to me all the body or network are 20 times less.

D

Sorry I did not catch. The whole comment of yours, I mean so. The thing with Randall is that it's also right now actively tried and experimented with in ipfs projects. So I've seen like issues and pull requests that people are trying to do that.

D

So it's it's already happening and, yes even said the whole after you, experiments are also toward sleepy to pee, so yeah I mean I think that it's totally probable that this code, like maybe this, protocol something different but but based on Mandeville, will replace the current discovery protocol and will be like a default before this cover appears discovery protocol- it's it's still, I guess undecided, but yeah I admit might happen.

D

So if we like have a good experience with it, we just collaborate, write some blog post like try to describe it, and let people understand that maybe a huge benefit in that moment. For me, I was.

A

Like yes, I'm just gonna suggest, maybe maybe we can identify other teams where this would be granted clients and so on, where this will be useful and just excel in chat like this, all core devs called to gauge the appetite of Adak light. If them commitment is to go with this, so it might be useful to reach out to other teams and in different it's all core, devs or I. Don't know if there's like the India fear and chat to something like this and see what are the team signal, or what do you guys think.

M

Yeah absolutely absolutely, and can we can we start? Maybe I don't know me tree, maybe you you the best person to do this? Can we start just by creating a table with comparison of 2, D, 3, 5 and and renderable, and explaining the pros and cons of both protocols? Can we can we start in there.

L

D

Yeah I mean if it was already posted, like Dmitry posted something like that, but like.

I

You like it what.

D

Table and comprised of numbers between this cover version, fight and rendezvous, but yeah yeah.

M

It was just numbers, it was just numbers on the traffic usage and it definitely random. Who is much better, but you know, maybe not so much, maybe maybe different kind.

N

M

Suitable for other clients, yeah.

D

Yeah yeah, yeah I think it's really good material material for for more research and for a blog post, even an Ori IP yeah whatever, but I think that the first stage would be to actually start using this, because we are testing that kind of simulated environment. We haven't been really tested that with with mobile devices.

D

Yet so as soon as we push that push it to the staging fleet, we could start like testing it with some some some real scenarios and later use it would be as a default and if you discover a prodigal and after that, I believe that it would be really proof that some at least in some some load how it operates and how how good it is so yeah.

M

D

Cool yeah and yeah. That would be it so so yeah, that's the state of pretty pure stuff right now,.

A

Should we move on to testing or does someone else have something else right, cool test it so.

H

I believe we should speak about internal tests. Right of tomato Testament was part of infra, and a lot of several weeks we've been focused on reliability and increasing coverage and decent progress and said.

H

Basically, we have two type of runs now extended run against nightly, build which have more than hundreds of tests and in those builds we have ten percent of average failures, which is great, and another type of built is built against PR, and we are in muslim half of 100 tests there and we have 16%, so we have less tests, but more trailers and I think that our future goal is to start implementing working called merge if tests are failing, compare and I want to like talk about that extended here, because we have a lot of people right now and I can get your opinions.

H

We made a survey before and only one person was against blocking it PR, which is great, also and I, just want to explain how it's going to work like steps fast. Oh that's to be done before. Implementing that block, like first of all, I I, want to explain four types of results that we have in automated tests right now, it's like obvious result, it's bad, but it can be known on you and average is 50. Percent of failed tests is due to box real issues and product known on you.

H

Issues also help feature changes in product and it's only 4% when Samsung changed works in new way or just disabled.

H

Twenty twelve percent for infrastructure issues and twenty eight percent for false fill up sites when something went wrong can test. Also, we have one significant issue like very important. Some peers have no end to end test results from what I can see. There were no builds against ours, like twenty percent of cars have the tissues and a lot of them are from extended contributors, so we are going to solve it nearest future and so how we are going to block we are from March. Basically, what should we do before doing that?

H

First of all, we need to fix all known issues from developed I'm going to make a link and chart are only for known issues which blocks which are blocking tests and saying, develop and also after we are going to add ability to understand whether it's a new issue introduced in this PR or it's like no nation. And what are we going to do? We are going to increase details on our error messages and we will have more detailed error messages and we are going to match results from nightly.

H

Develop builds with results in cars and if they are same, we are not going to markets like file but like born into the same result like the same issue is present in nightly built and such kind such issues. One block that we are from match so for infrastructure failures will just increase number of rerun attempts and it should define all right now. We have like Fiji changed and fall spoilers left to be solved before blocking appear at as we tried for a month or more right.

H

Now we are fixed into those failures and scope on the developers. Yeah turn it in the same color and reruns of tests. It works really great and but anyway like something can go wrong kind of wrong and I, don't know. Maybe we will lost all our infrastructure or we will get some big feature changes. So people have a lot of failed tests and for that we are going to add an option to unblock it, nor afraid they are. It should be available only or testers sighs science.

H

Automation is basically regression testing. If you have no results of tomato paste, it means that whatever and all the regression testing against this PR. So in that way, we have test, runs for each PR and test results, the tool for chase and what we are going to do. We just want to add for specific test run for specific barrier. Granted pattern unblock merge, so we will be able to unblock this if something went wrong, that's it and if you have any questions from objections, suggestion suggestions, much appreciated.

A

I would like to tell check with guess maybe Julian and maybe Patrick Pedro I, like what do you think is needed for in order to rely on our automated stuff, like what yeah? What's missing right now from your point of view as a developer,.

B

To be honest, so I don't have know.

M

Maybe it's the two weeks vacation but I. Don't have a strong opinion right now about that I don't know about the others.

O

Yeah missing from whether I didn't catch exactly what you said. No.

A

Cuz we talked about automated testing some, so I was just wondering what your thoughts are on this plan and if you have any concerns like what would be needed for in order to have trust stall made tests and so on, because I know in the past we've had like they haven't been hundred percent, as are they so yeah.

O

Yeah we're I guess the biggest point is indeed the trust and visibility and yet, to be honest, I think most of the closure developer. At least they just enjoy everything they don't even look at it. So.

M

M

One thing I remember seeing as an issue was that we it wasn't clear when something failed on your branch, or it was already failing in the parent branch, so it once we have this.

M

You know when it's failing really because there was a comparison between the parent branch and your branch and new tests started failing. You know, for me, that's a great step forward, because, right now we have no idea yeah.

H

We are definitely going towards that, so we want, but what Julian? What's the position? Why don't why there's a not Reutimann thought those results can like, except knowing that it's in develop or cars, any other problems like what should be improved. What should be changed for us to rely.

O

Well, maybe it's just about perception, but mr. Akeley we had a lot of false, positive and test failing, even if they weren't we're not supposed to fail. So maybe it's just that now those those issues have been fixed and we can start relying again on tests. But it's just you know from a psychological perspective. We we just get used to in your all those failures and so I guess we have to maybe synchronize and make sure we can. We can leverage them again to.

A

Try as an example, I looked at the last pull request. I was built two minutes ago in 95, 96 percent of intent has has failed, have passed, and this is just adding like a single constant. So that means that developers contrast them because there's two random tests that are failing Leitrim. What I can tell so I guess: I. Guess it's getting rid of these sort of false positives. That would be most needed cuz. Otherwise we can't block on much and I. Think that's that's what you're only saying it like you.

A

If you're developing you look at it and you see ninety something percent passed and then you know that the failed ones are just random failures. Then you you, don't trust. It I think that's the problem. It may be it's partly a inflection problem like maybe this is related, timeouts and so on. Maybe there's some need for instruction investment, I'm, not sure.

E

But when you look and right now the two failed tests how you know that it's it's not actual problems found like right now. If you are not blocking the merge, then we are not changing the culture like okay I can ignore it and I don't care about the tests, but the tests are there to actually to prevent anyone manually, testing it and like, as Anton mentioned, we sent a survey asking exactly the same questions like guys. Why you don't trust tests? What should be done and I can from Testim to touch Hydra.

A

I mean, in my view, like, if you make a pull request that change it's like a constant and it doesn't sort of break things then like you, should have hundreds and end-to-end tests, and if the thing is that it's a bug, so someone else has broke in sort of the test and, and that should be probably be priority to fix it. But right now it seems like sometimes they fail.

A

This is a random or it's or if there's no process around fixing them, because there's timeouts and all the stuff so I guess, like your point, this I should be hard to be sound like pretty much all the time. From my point of view, I don't know there.

H

Are slightly issues, for example, like you can open that link, and you can see that two of those review throat issues are happening from time to time in different transaction tests like you can see a replacement transaction on this right and.

A

So it seriously that's what we need to fix right, because until we'll fix that developers won't.

E

Yeah exactly and this.

A

E

An ochre and it's waiting for you to fix them like it's waiting for all developers to fix them and I'm, not sure like. Maybe we can agree on the better process, but yeah. It's it's in the github at least our test. Automation team should constantly remind about them, but it's it's like yes, issues, errors in it investigations looks as everything is there. I just need this.

A

Equation we can take care of in the mobile team because it seems like it's something that might shouldn't block releases right or maybe the test should be removed if they're not very useful. One of the two seems like.

C

It's an actual issue with transactions that replace meet reduction both created instead of a new transaction. That means we don't generate the right norms on the client. That means we have a bug that not you produced always, but if that's why they fail, they should fail for them. Right is.

F

There any is there any oversight on on tests failing I guess something does fail to do is there's someone that is the person who tries to submit the PR just ignore them or is there? Someone has to check to make sure this is a real bug or it's random. I.

C

Just wanna go check it's the bots in in the client right, so it creates a transaction that cannot be published somewhere because it is a record, unquote replacement transaction instead of a real connection. Right.

H

Yes, I can link that the issue like, therefore, each of which might be or might not be reproduced with hundred percent rate. They are linked. ah Yeah. Give me one second to relink, so.

H

Like we have not only the replacement of transaction but services take too long proceed from pending status and everything is there, and we can't ignore that, like the counter disables all steps, because we wanted the functionality. This.

A

Is something we can address as part of the release pipeline in the mobile team? What do you think.

C

The first one is definitely she.

C

Replacement transaction underpriced. It means that most what was generated. It is not in sync, with whatever was Sunday, so it's either India or our interaction with inferior is wrong. So it's not a mobile team. Really, let us go the second one, though we can just send it one more time. Most cases it doesn't work, one I don't see yeah. The third one is definitely mobile team.

C

A

It seems like what we need is. We need like an concentrated effort on getting it 200%, and then it should stay a hundred percent right. So.

C

Before this four are not like it at this for a flaky status so that that's what not not that amazing developer should fix that, did she, like total developer, should keep them yeah? We need to prioritize what.

A

What can we do to fix them once and for all, because we want to get to the point where we can actually make it part of and make it part of our pipeline. So we can yeah.

A

Testers, know that this stuff, that these things are taken care of and developers, can trust that the tests that they didn't introduce faults so any folds around how we can get that get that stage. Oh.

C

I mean work interlocking pull request might be good solution, even though some developers will suffer if not those who introduce these issues but well its benefit they will portion before system. Maybe we can well maybe.

A

We can have a spot of the next bug fixing day or week or whatever, where we just watch the Sphinx once and for all and then, with the end goal of it being hundred percent automated automated tests succeeding, and then we can turn off turn on that they are blocking. What do people think about that?.

E

Because I'd, rather not wait for any bug, fixing days that we have one god, I, don't know since like two or three months, but put them high priority issues and treat them like the most important things to fix after release. Blockers are fixed really like the control lines. That's why we developed new functionality or fix new box. If we have like high priority existing box, I agree.

F

C

I think we should do.

C

Yes, Adam, do you remember we have this issue before with the replacement transaction on the price once the go pipeline also must fail because of that every now and then I don't know, did we do anything to fix it or.

D

Oh yeah I would need to have a like a specific issue that that you're talking about because yeah there was a couple of related issues, but yeah I, don't know we were trying to fix them, but sometimes there's like just.

A

D

A

Maybe needs to try, we keep, we can focus it's like, but but it's just a way. We look at all their blockers and we make sure that they are assigned at least to a team and then maybe- and it would make sense if you can arrange like a another bug fixing day or repaired. I don't know. But but we can. We can look at it after call or simultaneously and just make sure that they all decide what do people think of that.

C

Yeah yeah just put them into high priority and fix them up. There is Orchestra fix what it is, investigate them yeah. We we need to.

A

Just need to be assigned to specific team morale, because that's the thing is it's falling between chess right.

C

White coats, replacement reduction, one smoking, so I scored, replaced. Introduction. We generate non on the for every transaction on the client side, so first we get it from impure ideas. But then, if we sell multiple transaction, we keep internal counter of them and the only way of from like there is this thing about your placement reduction. Elysium is genuine, so you send one transaction, but then you realize it's wrong.

C

You want to replace it while it's still pending, so you just send the transaction with different parameters but the same norms, and then the server is supposed to replace it inside pending. But the thing is with these transactions: you need to pay more gas for them, then to the normal transaction and yeah, essentially because of some box somewhere in our code we send multiple transactions with the same ones to insurer, and so it things that they are replacement transactions, but we still keep the gas price as for normal transactions, that's why it fails yeah.

C

Chuey we have talked about it separately,.

A

Should we move on to find out it's well.

C

I'm not wanted to say yes.

E

Sorry, just this summary after the meeting we can go to select which exit Oh interest and to end test workers are assigned to the team and get the high priority issues and get confirmation from each team lead and they will be treated as a high priority issues. Right.

C

E

C

E

A

Cool I guess client at its bigg or mobile first yeah.

C

So what here last week, we'll get security update and this week it turns out that there's India right and they have their own token or bounties networks have their own token there, and this token, unfortunately, is only available wrinkly, and that means that first read this token, which is a pretty trivial fact. But since it's a wrinkly and all our size, tokens were always on robbed them, so we only tested it as I understand on from Rob's 10-ton main net. So we kindly need to test everything in me. My queen can be wife.

C

So that's why we're just taking the previous release. So it's like we want to have absolutely the smaller regression scope with little cancer to the smallest one. So we just take the previous release and then just add like what it was like to commit to be able to to have this token token, for it India in there, and then we also add some bug fixes like to make green could be like right now. The problem is the Trinity.

C

We cannot really talk to any other networks because we keep winking across the separate so to fix that. So going to be, users can also chat with other users and status and after that we'll release. This will release it not like. Usually we release on Monday. We should release it by Friday, because Friday is the first day of the conference. So it's a little bit weird week in terms of repeaters. So that's why we won't read it yet.

C

Any other changes we'll take the previous week's release make some super minor changes to it. Then we release it just to be able to do this in time for the best for the next week. We hope that we'll finally get all the new updating instead we'll go the version that Adam was preparing and probably I, don't know. Maybe the new fleet that kappab wanted to, but this will be next week. Is it bi-weekly meeting I guess make sense to think about the next? One is wrong, so yeah I guess that's it from the mobile team.

C

E

Made a small optical, so they also would like to run some. What that's right on this.

C

To just be stimulating, yeah, yeah, I, guess what there's like 500 users and so far, that's what we know about it right. So we need to think it a bit more yeah we're.

E

So looking for the use case, but we also like checking right now, testing like getting the build ready to be tested for functional issues and lives there. No and we can actually communicate and rinca by and on these other networks. And if there are no more showstoppers of workers found. Then they'd like to go to law tests and by that time we hope to clarify the exact use cases to toward make sure.

E

A

Max with Max on desktop.

J

Yes, so recently we have discusses that we are going to have nightly, builds for desktop Sankyo Cooper for walking and looking into that, and once we will be zero, we supposin to start chattin internally in desktop app.

J

Also at the moment we have issues with Jenkins built for desktop it, because on Linux machines was not the much space and desktop builds, takes pretty low spaces. After that will be resolved. Then every PR for status react will trigger. Jenkins builds both from a boil and desktop.

J

This might affect everyone in general. Right now we still working on resolving some blockers from on low level from greg- desktop library. Is there like one two one or two serious blockers who once they will be result? Application should be available for in internal usage and should work good enough, a disease and also we are working on finishing different features for a user interface like push notifications and the audition interface, the result for different UI issues, so that.

A

Cool I just want to take a second trust people, so the corn inclusive, was it set for time got started. Maybe 15 minutes like 15 minutes to go from the scheduled time. We have quite a bit left to go through what people prefer to finish in 15 minutes or two people. Would people be alright, extending it and obviously people can leave whenever they feel like it on use for them and very preference like roughly? What do people think.

A

So either sort of thing.

F

I'm gonna stay.

A

Any other pings and.

C

Girls all right.

A

All right, it seems like we are okay with extending and if someone wants to leave, yeah just go ahead and leave, and you can also just look at the stream afterwards, all right next, one or I guess do we have any questions on client work.

A

Who research at clips where one thing could be a whisper, no densification and then I, don't know if that's a core record or someone else has something in her past. But we can start with Demetri and whisper note into the vegetation.

L

Okay, so so yes, I started doing research about that can be used for incentivized, nispom notes and at the moment I collected collected small documents. We shape a stack up of times and different channels, but but here is it again in the chat, so I guess I, don't have a I, don't have time right now to share all the details but made it make sense to go over main points.

L

That makes sense to go over main points.

D

A

Yeah, just just briefly to give people an idea, maybe also.

L

In cured, construction, I I propose to to make based on the reputation system, and there is a very good amount of research about distributed reputation, systems that work reliably in again distributed environment and can withstand known attacks, see goes fight ocean, some others. We should describe it there so so.

L

The first goal is to actually build such such system and prove that we can get reliable regulating from from the network or bidden okay, so so the first goal is to build such such a system and prove that we can get reliable rating from from it and then, based on this also straitened peers will be able to select other peers to that half way. Mobile mobile peers to be able to select pieces, high rating and pay them for service, so there is also research required to figure out the best.

L

The best payment model at the moment, I arrived at conclusion that it can be done over a regular payment. Channel and payments are incremental so like. If you wait for 10 minutes and then pay for 20 minutes, only one payment will be executed on Shane and yeah, so exact.

L

This I see: okay, I, don't I, don't figure out rose after The, Cove I probably broke broke, something yesterday changing things. I.

A

Guess that one question or what around this of micropayments approach, because it makes sense abstraction for like computer computer networks with us as humans? It's it's not Sophie nice! It's really not the main concern to have right now by own whiskers. How people think about this, because this paper by Nick Szabo and like mental transaction costs and the fact that micro transactions are not very useful for humans.

A

Generally, you want to have sort of higher threshold, so I'm wondering what people spot around the possibility of having like a higher level kind of thing on top of it yeah. So.

L

There is actually one important technical reason, a security reason for doing small payments, and this because we cannot, you cannot guarantee the rating, will not change over time and the fear and the fear which we will connect. This is perfectly trustful to trust us.

L

Perfect or imperfect here is perfect history and perfect quality of service, so we cannot guarantee that it will stay away exist next day or next month. So that's a reason to to make such new payments and another argument that there is one at least at least one. We were very good white papers that I found it talks about incentivized tor network. It's called orchid network. You can find that if you want and they also using such small small micro payments, it probably makes sense to show something different to using like, like.

A

Yes, we need to figure out throughput, but I maybe have more collaboration with you weeks. Research and stuff because there are like, like the phone cell uses, is one thing where people like you. There are ways to figure out, figured out and same thing with gas when you have a car and stuff you top up and like all these things. So, but maybe that's like a layer above and we can figure out when you exceed suits later.

I

One thing that the swarm team is doing is that they're a line service providers to put a state to promise that they make and if you can prove that they did not have hold that promise you get rewarded for it. So, instead of having small little payments.

I

Provider promises to be good and if they're, if they're banding, oh.

L

F

Yeah, how can you be rules, bad behavior, in the scenario yeah.

L

And it's exactly what we cannot do so as warm as I understand you when you store some chunk in some servers this server. Will you provide you with a proof of proof proof of conclusion to a common miracle? Erp directly pass assignment is visited several processes and then you can go and then you can go and prove bad behavior. If this server doesn't doesn't store this chunk some sometime in the future like we cannot do it. The only thing that we can provide is exists.

L

Approximation of quality and make sense, but we can guarantee that this is mrs. Pirus good and all the spirits bad. We cannot only provide the approximate rating for the spear and.

F

Those should be falsified, I'd. Imagine if you don't have a heart at a heart metric for proving bad behavior, that you can also falsify the claims of a behavior.

I

L

I

Is that message? Delivery can be verified like the quality of the message delivery system, for example,.

L

I'm not sure that's the case in this per would.

L

Proposals that they have for this initial written guessing that, actually you generate a message and send it to one peer and then because of the different properties, you must receive it from other peers yeah. So that would be forgetting because forget an initial feedback about each concrete pure. But you can broader terms, message: delivery. What doesn't mean only spur, but what is message anyway?

L

I

Yeah I guess there are some open questions with that approach as well.

A

It's a plane and something we can should look into more as well in terms of if there are ways we can get this to guarantee three alene I system, if you haven't read it, this is a swap, swear, swindle paper and it's really nice for the stake in solution and how he sort of takes guarantees. And then you slack, you can do slashing and partial, slashing and stuff, and it's yeah. It is kind of a similar programming thing to what's happening with if you're in point O.

A

So if we can regain some of those things, they would you're a nice property, but maybe it's difficult. We whisper I, don't know if things change with the PSS I, don't just like you have any thoughts on that.

A

If we're to move to beat us, for example, no.

I

It's the same thing.

A

I

When I would raise the point that, like this is this is similar to how how many contracts work in the real world all right, like you, set up a contract, that's what that's! What it's inspired by! Basically, you set up a contract. You make a claim.

I

Somebody else proves that is wrong and there's like a judge process that and that revokes that that sort of deals with that and and and the way they imagine it- is that if you can't have a reliable, a fully reliable system, you use an Oracle instead, and that Oracle is something that you'd have to agree upon with the service provider and with the client and you'd have to come up with a sufficiently targeted metric.

A

I

We couldn't happen.

A

Yeah, so maybe there's ways we can do it that wouldn't be perfect, but something similar well, you can use this. Can someone say that this is like a bad provider in some way and if it passes on fresh roll or whatever? Maybe maybe you can do things like that? Even it is not perfected, which so regain some of these copies.

I

Yeah and it's the same thing with the payment system is Rob's our swindle. They don't require exact tit-for-tat accounting. They they allow some slack. The idea is that, in order to minimize the overhead you your lot, you, for example, you might provide a little bit more service than then you should, assuming that it works in most cases and then bad behavior is analyzed afterwards.

I

Let's go sort of an optimistic model for service.

L

And also choose it become useless, not tit for tat, but what's a triple s, you use them for for mystery. It also doesn't I, don't see how it's useful, no case.

L

As are they going to use it for PSS for wick? Was there any discussion about it?.

I

Yeah I think the plants used four pieces to plant used for storage first, but then business as well.

L

Because you actually well, you know it's not the broadcast, you, your storage chunk and you retire. The chunk, so you're kinda explicitly state what your what your action is, and these pearl is just a network that this traffic, that is just broadcast everywhere.

I

Yeah, but but for example, for a mobile phone, but it could be used for is that we have super whisper nodes which provide better proof of state, for example, for messages that get forwarded that way for payment right, and that is something that is easy to prove that the message got sent out in the network with the given proof of work. That's something that you can verify by room by running it. Yet another client right.

I

If you suspect that a server is not forwarding your messages, you can take specific action in order to measure it and then report the results.

L

Yeah but again.

L

Some more code, the monthly trusted in this case, because you can like.

A

What do you guys think what look for keenness to discuss post and then because there's lots of research paper and stuff and lots of things to think about. So maybe it's a.

I

Start idea- and this is just an it's just something that I think could fit, but I haven't looked into specifics. Oh sure,.

A

And yeah, yeah and I guess: there's also like one of the things we want to try to do. This quarter like to figure out this, this kind of plan and either like something I would do me just saying. Let's make maybe there's other things as well, that we can do like high precision. So it's not something we have to rush well, so I know Ricardo or Perry or other smart connectors may be just some.

A

Some thinking you have, or whatever things you've read that are relevant, is the useful tells many people brains as possible on this three.

A

You just like you have any any other research subjects.

I

Nothing more right now.

A

You so I guess next one product updates, if you're anything specific under water to that so I guess we're going to talk about PFS after but I guess is there anything else around chat.

A

You just come back for maybe or wallet corn.

K

Hi well, there's nothing really super interesting about wallet at the moment it's in bug-fixing with in cleanup phase, but so you know right now, I'm factoring the wallet namespaces. Basically, the idea is to make it more testable with unit tests right now, we'll have more pure functions, ideally, no so to reduce the regression surface. I guess the one interesting thing is the addition of ENS user names in invalid sent. That's currently like about to become WIP.

K

So there's that and that's actually that's actually the thing we reused for security champions experiments as well for the threat modeling. So would it be interesting, yeah, eager I know this is an experiment, but maybe we should like publish the end results or something for others to you know we should definitely use their findings.

C

Yeah I think, first of all, everything every thread: fine, due to the threat, modeling I think we should publish- and we shouldn't probably compile some kind of reports on upon this- how it went and correct.

J

C

Think it will help everyone to be on the same page but yeah as soon as we get easiest thing, I thought, I guess I can just do into the report and share the dockings maybe end of week.

K

Yeah we're visited India thing anyway, yeah the token that yeah, so, basically, the bulk of the new things in in wallet is really.

K

It is really on the refactoring side, which is actually nice change after the the bug-fixing yeah. Does anyone have any like questions around the wallet area for me or anyone else? I.

F

Have some personal but I think that's gonna, be with just that one on one that I have later on yeah.

M

A

Cool yeah this under here.

N

A

That's right: oh yeah go ahead! If you have something good that no.

N

No sorry so you sit in there tell me next time all right, cool.

A

But I guess yeah it's it's! You and Rashad Yeah right, Esso, andrea and check that stuff.

N

A

Exactly exactly because I'm here for death, I, don't think there's no one else on that today. Yes, sir.

A

Yes, so please please undergo regret. Okay,.

N

So I can give an update of where we are at the moment, and so that's probably no. We are implementing DFS so far. We, the strategy, we're gonna follow, is basically like follow quite closely what signal has so XP d h, plus double action.

N

So the moment we implemented a basic XP, the age so that the clients can share a key exchange structure, key and communicate with Rita, and we have implemented persistence using SQLite encrypted in status quo. So we can store key and safely on the safe side and now we're moving on to use different approaches, we're now implementing their self.

N

We put an implementation that was done for States some time ago, so just a matter of making sure that it works at the moment there are some tax, failing as you need to in today, and you know just making it compatible, so they use a different bees, a different key. We use helium key, which is like CP 2 5 6 k. They use two five, five, zero, nine keys of Canada, and so we need to sort of like make them talk to each other.

N

So that's basically the applicator DFS there's some discussion and discuss on basically how to solve the initial key exchange, which is the outer sphere and that we are not quite sure yet how to solve, and so that's that part hasn't really been. We don't have a strategy, yet we have few ideas you can followed. You know it's up. Basically, that is to essentially requires some sort sort of storage.

N

The idea is to use anything that we have I know like right now, so we can use it. For example, we start with you know, propagating the band of content, updates, public charts, QR codes contact codes, and then we move on to DNS, because your nephew can apparently you need to. We need to make sure that that is the case. You can store arbitrary it sub data on ENS.

N

So that's one of the way we can show if I'm gonna move to other suffered from such wall, which is don't read it on yeah now, and so it was like. You know they keep mood after runner or glass that do that. But then comes you know not just now. Let's say that's an excitation.

N

Yeah, that's pretty much. It.

A

I have a question on if we're moving things to, let's go side mm implications on desktop and I know. There was some stuff around our missus passing works and some some bugs and stuff like I, see some cognition that me it's neither there I know max I mean.

N

Basically, like status track is, very you know this very different changes that are required to make this work is very, very transparent so, like instead of you, just use a different end point. When you make a whisper call, you don't go whispered directly anymore, so you don't post, you just make a different call. So that's all the changes that are needed in on status track, so I would think that works with desktop as well.

A

What cool? Yes, there are new out of forks rounds of the initial key exchange like it's something to go underwater.

A

All right, Andrea, whisper wallet, key idea, yeah.

M

So so we've been thinking about this for a while. So now we started an idea which is working pelvis and there's a topic in discuss so just work up. So the basic idea is to to use two different keys, key pairs for whisper and for the wallet so for now we just have one. So it means that the public key that is used for the chart is the one that is basically using the derive the address in for the war look.

M

So the problem is that, just by charting anyone is like exposing their wallet and all their transactions, especially in public chart where any active user are like exposing their identity.

M

So for these reasons, so for security and privacy, we want to like extend the pistol that we have now to have two different keys, one for whisper and one for the wallet, so that, if one is compromised, is not going to compromise the other one and of course there are some compatibility issues under it, but also the other, and they are proposed from some things that you can read about in in discuss.

M

But in general the first thing that we we need to do is add: Jing changing the protocol so that, before sending some value through the chart, we request basically the the other user address, because because the current implementation is like just using the the con, the chat contact and then deriving the dead rise in this case will be a different different one. And we can maybe release the first update with dates. And then we can. We can completely change the the application to use the two different keys and I. Think for sure.

M

We will need a lot of UX and UI work on this, because it will be a longer process when you wanna start sending some value in the chart and also from the security perspective. We need to maybe talk more about it and see if there are any issues.

M

Is there any question.

A

Do people have any concerns about this? I was.

F

Curious about whether or not these two, these two keys when they're, separated or derived from a single seed, Frasor, there's gonna be two different seed phrases associated with them, or is there even a seed phrases? Okay,.

M

Yeah, so for now, so the idea would be I mean this. My idea would be to use a different elevation path from the same scene so that it's yeah, so that one account is recovered or created with the same mnemonic phrase, and then we derive two different keys, but I mean we can discuss about it just understand. If we, if there's any other I, don't know security concerns or something yeah.

J

F

I

Doing at the same by itself,.

M

A

All right, cool, I guess before we should finish up with some meta things, it's just something that people feel like. We should have talked about that. We haven't.

A

And maybe something around less so you'll see or some stuff around contracts and ow stuff or anything with it's.

A

Alright cooked so just briefly, some to this. This is obviously first time we do this so I guess any feedback like in generally speaking was this. Was this type of meaning a useful.

F

To people for me it was incredibly useful because it helps me get up to date, what's going on quickly, if, if possible, those who contributed, if you could add in to the hack MD, if it's not, if there's something you needed to say or said that wasn't included in the and the notes, can you edit yourself, yeah.

M

I was going to bring up the same thing, make sense, yeah.

A

Plus one is so we're gonna merge this and then it will be part of the git repo, so people can refer to it later. The others suggest.

I

A

I

Kid time, I think perhaps an idea is to focus on things that need a decision or are about the future in this call, because we already have the townhall for things that are from the past, for updates and of things that have happened.

A

What time so, what can improve.

I

Yeah, just just a general note that if you, if you're bringing an issue to this particular College, perhaps it's good. If it's about the future and not so much about updates on what has happened already.

I

Just as a way to manage time because we ran over by whatever.

A

A

Cool what did people particularly like or get out if it wasn't some here, what particular useful to people so.

E

It was good I know, I was late, but may also add to the previous question. So I completely agree with the structure in time and visit genders, because it's not only to talk. We need to understand what we need out of each topic like. Is it a presentation how it works right now and then like proposition and then what actually is needed like solution or agreement or next meeting or anything?

E

And of course it's the first one, so you never know and I really appreciate having something to watch when someone is talking because sometimes I can't follow. Sometimes there is interruption on the wine and it's it's not easy. So if something can be like, you can share the screen and show a bit of thing you're talking about. Oh, that would be helpful like this to chance. For me, that's it.

A

A

A

A

All right so yeah I guess if it would be very useful, like if you like the sort of look through the notes and if you feel like it can trip something will be great and later this week push it up to the git repo and then we'll have another meeting in two weeks time, and it will be an agenda like a crate like a scaffolding issue where people can add stuff to places, add stuff that you want to talk about and what you want to get out of it.

A

And then, if the recording went well, we'll have some kind of stream and put it up as well and then for next meeting, we'll, hopefully have slightly more robust solution. For that, maybe me, but it seems maybe assume or something, but it seems to have worked quite alright, we'll see cool. So thanks much for joining everyone.