IPFS GUI and Browsers Weekly, 4 Mar 2020

Previous Meeting Next Meeting

⏯

youtube image

►

From YouTube: Metrics, users, EOY goals and pausing window.ipfs 😬 - IPFS GUI and Browsers Weekly 2020-03-04

Description

Meeting Notes: https://github.com/ipfs/team-mgmt/pull/1123/files

About IPFS GUI and Browsers Weekly: https://github.com/ipfs/team-mgmt/issues/790
IPFS Mirror: https://ipfs.io/ipfs/bafybeicx2rf7f6if7feeqt3cf7pnyoywony2cu5g6dcqkoloatvqyzppue

For more information on IPFS
- visit the project website: https://ipfs.io
- or follow IPFS on Twitter: https://twitter.com/IPFS

Sign up to get IPFS news, including releases, ecosystem updates, and community announcements in your inbox, each Tuesday: http://eepurl.com/gL2Pi5

A

Welcome to GUI and in web browsers for the 4th of March 2020 I'm joined by regular folks. Those are new regulars guys.

A

Now, let's take a look at our agenda. I've had it yeah I think I've added the first one, so I can start and then feel free to add anything else. We should discuss this week so the item I've had it is about window.I TFS and the fact the background is that jeaious ipfs and JS ipfs HTTP client both switched to Assange iterables, which changed the programmatic API that people use in JavaScript if they switch to the latest JFS and.

A

Corresponding HTTP client, the problem is a keifa's companion is exposing a certain experiment. We are exposing a.

A

Programmatic interface on window object called ID FS and you, if you have IP first companion installed, it will inject that API endpoint to a page and a page can request activation for that interface for a specific commands and what bastes? What basically happens behind the scenes? There's like a proxy between a web page and the API PFS api used by ipfs companion that could be embedded, GSI BFS running in a browser, extension or remote HTTP api that could be localhost api, provided by go ipfs or ipfs desktop.

A

So on some level we don't really control the version of that API that IP FS companion connects to generally ipfs companion has some fixes that make when we started when we historically had the breaking changes to files api's some commands got moved from files API like ipfs files, namespace to the top-level namespace we've added aliases. So no matter which version of API is in the background the programmatic interface exposed the web page would still work as expected.

A

The problem is now: we made even more breaking changes with that a single factor, and now it's no longer like possible to programmatically detect what like the web page wants.

A

Web page may expect the old version of a TFS cards and now or it may expect a new version which should return as an iterable like reported on this issue, but it was not because ipfs compile is still using the old version of JSI TFS, HTTP client and that client is exposing the old programmatic interface.

A

That deep long background was like necessary to understand the question I post on this issue. So basically the question is: should we disable ipfs? We do ipfs until support for those new api slants and what I mean by support landing. Basically, we need to switch I'd, give us companion to New Jersey, DFS and JFS HTTP client, which used those new api's anyway. It's just. We, namely me, had no bandwidth in this quarter for doing that. So we postponed that to you two sort of there's like open issue for tracking that immigration.

A

If anyone is interested, the problem remains right now. If a website is using window ipfs and the way people are using window, IGFs object, they check if it's present and opportunistically they try to use it. If it fails for some reason, then they can like fall back to spamming their own boat or using like external HTTP api. However, like usually, if the window ipfs is present, the website tries to use it. So the idea is to right now it tries to use it and it breaks which is pretty bad.

A

So the idea is to push back histories of IPs companion that would simply stop exposing window ipfs and make it like grayed out blacked out in preferences, with a note that we are in the progress of in creating two new api. That's not like very severe change cause. I don't feel a lot of people rely like 100% on this. It was most like opportunistic. If someone is running a node, it's nice for people to use their local node instead of spamming, a J's IVFs for fetching data, but it's still a significant change.

A

The question is: should we do it I'm, open to opinions.

B

So could we like change instead of like disabling? Could that appear thing for like detection, where they check, like the ipfs version, could make that recommendation? Does.

A

The thing right now they, if you are the website developer, you can execute ipfs version command, so tell the version of the backend demon. However, we don't really have a way of checking the version of HD client or JSI PFS, so those are two different versions right, which expose the same core interface and I. Don't think it's a burden that we should push to developer because that would require like developer to know.

A

Oh those versions of J's, ipfs and GS IPF SFTP client exposed this version of programmatic interface, like we don't have any versioning of the programmatic interface actually like. We have the core interface, which is like released but I'm, not sure. If that's like an abstraction that we should like exposed to developers.

A

When we started window idea, firstly, experiments the API was basically like one-to-one with HTTP API without any bells and whistles, and the assumption was it's sort of like a static beast and it will check when we make a breaking change. That would be like a TI hito v1 and with that we'll probably introduce some versioning scheme. But right now, I don't have a means of doing.

C

Versioning, the core API with something called a specification, sounds like a great long-term plan to me.

A

Yeah because, like we talked like we mentioned before, I started, recording a lot of people rely on Jas and like Jas itself. This is this abstraction of programmatic interfaces effectively becoming a separate API that we need to like cater to and making the problem is in companion. We don't have any means and people don't check the daemon version when we made like a changes to the way like ipfs, odd hundred something people were not even able to very account a version of go idea fest for working around that. So this is even worse.

C

Yeah, when you say people can't don't check the daemon version, though, like who are these people are these people that are using cheese, I pay fast, see their HP, client or yep. Yes,.

A

C

To local demon, yes,.

A

So basically, when you use ipfs api, you've got a version command, but that does not do what. Probably people think that it does. It returns a version of the implementation that could be a version of coati of soo-ji PFS and it returns a version of protocol which is always the same like 0.1.

A

It does not return version of yeah. It does not return a new version that could be bound to that programmatic interface. I. Think, that's that so.

C

The version doesn't match whatever the set of API and the semantics of the API and behavior of the API yeah.

D

C

A

What's worse, you can point like older version of HTTP client and then you a sink iterator version of HTTP client to the same HTTP API and like the demon version, will be the same, but the Jas program, like the programmatic interface for J's, would be different, so yeah.

C

I mean I, secure it, but I think your idea about versioning the programmatic API really makes a lot of sense. Like that's the, if that's the one, that's the main compatibility point that developers are going to use, that's the place where there's gonna be problems if API differs or behavior changes. If a semantics say the same, so I think versioning, the programmatic API really makes a lot of sense and if, unless there's any changes, you know that you see they're gonna be big.

C

To that to that set of API is I mean it might just really resolve the issue for awhile. We we can depend on that bit, yeah so and then ipfs. You require ipfs too later on down the road yeah.

A

Like in the long term, we could like extend of just the version command. We got like the daemon. The protocol we could guard like add, like programmatic interface version. As a third thing, the problem is solving the problem at hand, which is companion, breaks, websites who rely on window ipfs, because the api changed yeah.

C

But do it so we know there are 25,000 installs of companion, yeah, alright and I guess. My my concern here is that we might be over rotating on a fairly small set of people that we probably have pretty high level communication with.

A

Yeah the question is: should maybe the question is not, should we disable or should we do anything until we refactor companion to use the new API? Maybe that's the other question yeah.

C

I think that's a more pragmatic question.

E

Just on things, so, if we disable it, it will break the apps as well right because no.

A

Actually, the way people use window idea fess is like opportunistic thing, so they check if it's present, if it's present, they will use it. If it's missing, they will spawn their own J's ipfs or they will run their own HTTP client to some remote API. So it won't like. If we do idea, first disappears that will not break ups.

C

Yet generally yeah, unless the codes just pretty bad but again, I feel like we're talking about still a pretty small number of people. Yeah. The problem is more. If we remove it and stuff breaks, then we'll know a little bit more about those people.

A

Yeah, so it's more about objects, cause like people who use IP fix, companion, probably the most passionate about a cubist project, and they are the people who will test those apps who probably have support this opportunistic support. So it's like I worry about this overlap. Yeah.

C

I'd really I'd want to dig into who those people aren't, though, like if they're the most passionate users and they're the ones who are debugging this, and these are the developers and then they're the ones who are also going to read our our blog posts. Where you talk about this change or the one we're gonna retweet about it or the forum where you post about it and I guess: I'm, hesitant, I'm, hesitant to make too many assumptions around levels of usage correlate those with number of installs and like.

C

If you seem to be referring to your specific implementations. Where you've seen people do this in their code, we should be able to detect that to some extent on github by doing repo and code searching I figure out with the list of projects are, that are say, looking for a window TFS in there and their JavaScript or web content.

A

You can define fallbacks and we do I give us this. One also like implicitly enabled I believe so we can check who depends on that.

C

Yeah, it sounds like removing it, even the cost of removing it according to you is just not gonna, be really that high. A bigger challenge is making sure that we have an adoption pathway, migration pathway that remains smooth. So what developers who are adopting now they'd have to do too many changes or jump through too many hoops moving forward.

A

Should remove it for now and then when we introduce it, we introduce it with like that versioning. So it's more or less for sure somehow does.

C

Assume have poles, okay, you can't you start a poll right now and then we can all there's.

A

A pause button was when I.

F

A

I press it then window opens and then I need to create the poll in my web browser. Oh really,.

D

Let me like create yes and action. No, this is show up in a chat. Should it be a multiple choice? No I will make it.

D

No I will make it multiple choice just for magic effect and.

C

And also I wonder if it's anonymized or can you see everybody else, vote okay. This is kind of derailer. Okay, definitely I think there's probably totally fine to remove widow, not a few thoughts for that. It's.

A

It's like did anything happened because I pressed something thanks: tomb, apologies to people who are watching this on YouTube and other places. We should do better next time. I will end this topic and, let's provisionally, say that yeah probably we should remove it for now and then we'll have a separate discussion about versioning programmatic interface in a way that developers can actually use.

A

A

Next, one functional configuration of J's to be.

B

Yeah, so in terms of like usability of live, p2p, we've been looking at like how we can make configuration easier. We talked about a myriad of options in last night. I had some thoughts that I wrote down in this issue on using functional configuration for ipfs, so I think there are a lot of potential wins to doing this, it solves should solve a lot of problems. It should allow us to export configurations as well as look at like what improvements we can make in terms of like actionable errors.

B

So if I do an NPM install the p2p and then I import, one of these configs from the Lib PDV package on a limb, p2p tech create because I want a browser package or I want a node package and then I don't install any other packages.

B

So I do you know, try to run that like getting actionable errors that say, hey run this NPM install commands to get all the packages that you're missing so kind of like guiding people for that, as well as like supporting users who want to do more complex configuration and I think this will also allow us to do things that we've been wanting to do for a while, in terms of like splitting up the DHT into putting it in content, routing and appear routing and to peer discovery like we'll, be able to split up those components and have that configuration function, live in the DHT repo, so, as modules are created, if they need any complex setup, they'd be able to just like export their own configuration functions for users to to do and compose the p2p.

B

So there is a proposal in that issue. That's linked in the notes. If you have comments or opinions, please please share.

A

This is super useful because I've been on the receiving end of the current version. Well of the current way we overwrite the p2p configuration in JSI, PFS and I. Think yeah right after you posted it. I linked changes, we've made in J's ipfs to make it a little bit less painful and I'm inactive as companion I'm overriding a lot of stuff in the default configuration for brave cause. That's a tricky tricky runtime, for which we had no template.

A

We have like a browser defaults and the node defaults, but brave is a browser with TCP transport and then DNS discovery. So we like I, had to basically write custom custom, configuration with custom list of transports and discovery methods, and it's like already bit better on the jsapi fest site, but I agree.

A

It's still tricky, especially like that.

A

Not sure if it's like not the railing too much, but the thing that always felt icky to me was the way WebSocket star. Webrtc is kind of fine. We only like past pure idea to to the constructor, but WebSocket star historically had a pretty bad additional steps in J's, ipfs related to basically like replacing multi others.we with WebSockets Darwins, not sure if it's something we can address in while we make it this refactor. But it certainly sounds like a problem.

A

Not only I would have because I imagine a lot of people who want to run a custom config. They probably want to run the custom configu if like something like WebSocket star or a custom list of transports, and they that's the first problem, the heat. So it's super useful to yeah, at least at least look at the that's a prior art for potential use. Yeah.

B

And so I think that, like the Jas J's ipfs configures, like that most complicated configure that we know today so I think if we can, we can build a configuration in a way that makes that just like ok. This just works now like we just have to add, like maybe specific stuff around, like listen and announce addresses, but that should also be done as like a customization function of here. They listen and addresses so that everything is just those specific functions and we should have like stuff just built out of the box to do that.

B

So, ideally, I would like even like the stuff that we need for the JSI PFS config. Most of that should be able to exist in like a prepackaged customization function in Lapita P. So then JSI p FS. We just need to import Lapita p, pass a specific configuration function and like maybe one or two other small functions and then just get the instance that way.

A

It probably will need to take another pass out your notes and revisit my mind from brave, because I feel I had like more pain points. If it's just like it's super quick.

B

Yeah I I need to do a pass at it as well. I wanted to get it out of my brain, because I woke up at 3:00 a.m. and for some reason thought about Limpy to be configuration so I had to jot down some things. Then go back to bed.

A

But also also going to sleep waking up or asleep.

A

Jaesi jaea sleep, p2p configuration.

A

E

Will also have a look at your proposal this week.

A

Yeah cause like yeah I I, wonder if required rocket Stardust would have the similar challenges with that additional code that had to be kinda run outside of the p2p and.

B

Really, none of that should need to exist anymore with the new version like we can see like if something like that still needs to happen, we can go back and look I'm like okay. What's what's going on, because even with the current configuration you shouldn't need to do that, because one of the things we did in the refactor is now we do. Whenever we instantiate transports, we pass Lib p2p anyway,.

A

B

Can go, get lit, be TP, pull Eid pull the addresses. Do anything they need to do.

A

Alright, should we review browser and connectivity metrics specifically, can we measure in browser notes in any way, should I share my screen with ya.

C

I wanted to just go through real, quick and validate these and see if this made sense, if they, if they were possible and asked for you all to add any notes there around where how we'd hook up the data source. So if there's a way to say programmatically do companion, installs that'll, add a link there to notes, for some of these other ones just add a link.

C

If you know, if there's any data sources available already and then we can hook them up in the grammar I'm, looking at adding refined plugins that make it as easy as possible. If rebuild. Let's do that without us having to you know, spend spend lifetimes and in the animals that we can't get back so reducing the pain of wiring. Some of these up, but I think that I'm, the the core.

C

The most important thing for me, is what they're, not any of these makes sense for measuring how we're getting to our goals in overall in browsers connectivity land. So what are the most important bits for us are and I think part of that comes to you know some of this understanding. There's there's specific shorter term goals like getting a bootstrap, no load decreased, getting relay users to zero or whatever, right or W a star that users to zero.

C

So things like that are shorter term, but those don't necessarily speak to the longer term impact that we want to have which has higher levels of connectivity for increasing numbers of people using IP FS from webpages. So I'd really I.

C

Need you all of your help in figuring out what the right way to understand and measure that, and um you know, I think all he was talking about the way or the stuff that are doing for like DHT crawling and things like that to be able to measure of size of the DHT and the ability to see have some view into number of in browser notes.

C

So if Jessica custards running in HTTP web pages are connecting and and connecting to other peers having a view into what what that means from is is, is that never growing? Is that number decreasing that there's probably some wrinkles around, like the works, that three boxes doing and I'll talk about that in the next one of these? Next items around how to collapse node usage across tabs, so you got one.

C

You got eighteen tabs open with some well somehow one jjs ipfs instantly time, some of the stuff that he was doing with lunette some of our partners, whoever having increasing deployment sizes like three box are really having to push hard on solving that problem, which of course then collapses. The number of users I. Guess it's still the same individual person. That's not such a big deal, but you know understanding what the numbers actually mean.

C

Knots don't the standpoint, but let me know if there's if any of these things seem like either they're not worth measuring or they are impossible to measure. Definitely let me know and then from that, like what are the? What are our what's our end goal, what does it look like when browsers and connectivity group is winning? Does that look like some number of connections from individual and users in browsers is actually increasing? Does it look like number of connection options increase.

A

If we had the HD, it will be easier because you just growl the DHD and you look at multi others.

A

Now we need to figure it out.

A

C

Alex said tht in the fall snow, I thought we had just agreed in in Brooklyn to not make a decision about tht in JavaScript and fall, but Alex said no he's like no. The idea was she'll just implement in the fall and that will be done. I was like okay I'm fine with that.

D

In the same room and everyone went.

A

Out from the room with the decision, they really liked I.

C

Thought my my understanding of the decision was conservative. I just thought we were telling it. We were pushing off the decision itself, I.

G

Pushed it off and then Alex decided.

C

I we did find out yesterday that three boxes using JSI TFS on the server as well, and it was an interesting conversation and we'll roll some of this up in the clouds report. But it was interesting in that they had some perceptions around capability of go versus, go I, confess versus JSI PFS. That seemed rooted a little bit in kind of the attitude of the project as a whole. Around like.

C

Oh, is the official implementation and I perceived sense of lack of support for certain things in JavaScript and not clear, read map from them in understanding what level of capability or support gsfs is going to get in the future. So I feel like there's like we're finding more more users of the software where we already make that we didn't know we're users of our software, even though they're very close collaborators and more viewing to that will be.

C

You know getting their feedback on things like whether or not the DHT support would make or break their use case, but would be good right now. They have a pretty constrained use case, so it it's not really a problem for them. It's actually really similar to what Microsoft is doing where most their clients just swarm, connect directly to them anyway. So not a big deal, but they want to they want. It actually wants you to grow and be more decentralized in a way that they don't need that as a as a crutch.

A

Yeah, absolutely and I sort of I had a call with terminal Monday, specifically around changes in the browser context and.

A

They probably will be pushing the boundaries that were visited by reckless lunette for a project. At some point, it feels like the problem of like multiple instances in multiple tabs or multiple repos repos pair origin. How do you share repo or like entire Jase I, give, as instance, across different websites that use the same like library or how do you bootstrap websites for just like having static page with CID, somewhere somewhere and a small library pretty pretty exciting?

A

Still the limitation of web browser context WebSockets web RTC, not working in service workers, problems like that, so we definitely continue talking with them, because they are still trying to figure it out what's possible and that will also feed back to the decision they make and we want to be at least acting as a guidance for what will be possible on the IFS site in the short term long term, when the DHD learns how useful it will be if every multi other is tcp one things like that, it.

C

Can is there, an issue are in? Are we distributed yet for WebRTC in service workers? Are we tracking that request in a way that all our partner projects can happen and make their needs? Ideally, yeah I believe there's.

A

Like wow, there are at least two issues about WebRTC. One is about sort of like a state of the art, and it's like all the issues. Not only that one and the separate one is about like WebRTC next generation, which is like of other quick and things like that. So dad like to two issues, I'm.

C

Just reckon I'm wondering, like you know, we hear three box talked about that so, like we've kind of jump to the next issue there so definitely take away before we hardcore switch away from metrics. Please review those metrics add any notes. You have around data sources for them and then we can.

C

We can work on getting those hooked up, but from the from the you know, I added an issue around both terminal and three box, because they're talking about these same problems, they're talking about things like multiple nodes inside a single browser when you get a bunch of tabs open, they're talking about connection limitations, things like web RTC and service workers came up in conversations with both of them, so making sure that we can use places like are we distributed yet as a multi, multi party way of people raising their hand and saying I need this, and then we can point standards groups to that when we go and talk to those groups and then I think also really even even envisioning I feel like there's a nascent opportunity for us to envision the non DHT ways of people that are connecting over WebRTC.

C

The way that meta mask is doing the way that web torn is doing a number of approaches for really building large client alene networks over WebRTC that we kind of haven't really been taking advantage of and I don't have enough visibility into why that is, or this is just lack of resources or experimentation, or we already evaluated these. We say these won't work is as alternative P transports that will run over WebRTC that are not the HD so, and people's thoughts on that would be really interesting.

A

I just remembered who goes experiments with DNS and the discussions we had on like shutting DHT on the DNS records.

C

D

Yeah, let's sleep on that one: how.

C

Come everything I bring up, you say we need to sleep on because.

D

This girl is just one. Our topic is another.

A

C

Keep bringing up little asks like please reinvent or throw away or burn the DHT down.

D

C

What's about to enjoy yeah, just so I linked the pl event issue there. You know both the research team resonate and us all pulled back from from IT or four to seven could just to close, but we do want to have a more concerted effort presence around 108.

C

We're kind of talking about things like running a social or a meetup like people are interested in this kind of stuff and have it be kind of like birds of a feather type of made up for 108, but we'll, probably because this group in particular it would be great to have your presence if you look through like the 107 schedule, obsessions there's just so many sessions that are relevant to us.

C

So I definitely maybe marked this period down on your calendar and put a hold just in case, because we'll probably want to combine a browser connectivity hack week along with with all or part of this meetup.

C

Hopefully, travel restrictions will be able to be less by that point and for a lot of you, this is some fairly close. Close travel, not too crazy. Just Jacob and I have a longer room to get over there. Nope.

B

I'm in Germany, oh right.

G

D

I'm on the wrong side, the origin Kevin.

A

This this site is much cooler and.

E

C

See us the next, the next one to you co-hosting, co-hosting, don't let the dream die way. I saw this thread from data hoarders on Reddit talking about suppression of coverage, 19 News in in China. One of the you know things that we have. Maybe we even talked to matters news would be interesting to talk about how we can support them, but co-hosting seems like we still would keep coming up with these.

C

These cases for co-hosting that seem like would be interesting experiments and how we can use the browser to be able to make it easy for people to publish from from I published from old web to IP fast every time. There's a situation like this, where there's the data suppression at scale, just even starting to happen in the US a little bit. It's really it's an opportunity for us to be able to say look at the power of these tools that that we built with co-hosting. We have some experimentation.

C

The last fall see like on top of them FS.

C

How can we build things on top of em FS, so they require protocol level changes that require API level or record level changes, but I'll, really like smooth the kind of user, onboarding and flow for being able to republish web content with my PFS and then read it and collect it back so I feel like we, we kind of set this we explicitly set this aside, but the use cases for for this type of functionality continue to continue to increase, especially this idea of like Auto, co-hosting, small bits here and there as you're browsing where people users opt in for a specific use case.

C

Maybe this is something we write up as a grant might be an interesting idea, but there's probably a little bit of work there around. You know testing and making sure that the the scenario actually like getting the the basic flow working first.

A

In case someone in case, someone is watching this and thinks I could maybe hack on this. There is ipfs shipyard co-hosting, which is just like a description of the experiment. Why are we making the experiment? There is a specification for co-hosting basic operations, adding removing listing everything is more sort of like a convent convention over configuration. So it's just like a set of convention for where you put snapshots, and how do you tell which snapshot is the latest one? How do you remove it?

A

This repo is just like a meta repo for the specification and overall overview where we are, but there's a separate repo called ipfs cohost, and it's a command line tool which implements this experimental spec. You can use it for like co-hosting specific websites.

A

So it's right now it's for entire websites, but it supports lazy mode in which you can you just or the sub pages that you actually visited or you actually care about, and what could someone try to hack on or contribute is submitting appear to give us companion, because this project is not only a common line tool. It should also act as a J's library so, like actually you don't need to implement the spec. Even the library is more or less probably done, because it's used by the command line tool.

A

Someone could try to use the same library and simply add user interface to a Kieffer's companion for like adding and removing sites or our browser action right now, in activist companion, there is just a big toggle for pinning, so you just pin entire Wikipedia hundreds of gigabytes, which is not the best idea. Well, um what one could do is to try to replace that that the interface for pinning with something smarter built on top of this library yep, would be really cool one of those projects that are like really cool, but there are.

C

The priorities yeah, it seems like if we could get the base pieces in place than other people, content verticals like event response and any specific thing like Wikimedia, we do things like people could run with it. If we got these base pieces in place, so maybe maybe I'll do something like maybe I'll make a calendar event for like ipfs dev grants hack day. So we just like set aside half of a half of a Friday or something like that to write up grants for things like this.

C

Yes, we didn't. We wanted to do that in a Bricklin, but we didn't really get time.

A

Yeah right to care grant takes time. I know: cuz I wrote one and I'm in the process of right. Oh I wrote two actually is.

D

There a leaderboard is there a prize. Yes, there should be I'm living, so I am asking about the potential.

D

There absolutely should be.

C

All right so I added this next item as well, and this is this is also a request for this group. Is you know at Brooklyn we got together. We talked a little bit about like the specific initiatives that we have in place. We we have. You know things like deprecating to be a star improving activity options.

C

We have some things around companion and we say have things around supporting our browser murder partners, but but what I really like seeing this kind of goes into planning for q2, okay, are so keeps use about him just a little under a month away.

C

So people should already be starting to think about what that next set of goals are, but what I would like to actually do is flip it around to me even just half like and I wanted to do this in Brooklyn, and we just really didn't have that the timer focus there's so much stuff going on, but I'd love to be able to have like where what the things like at the end of the year like they, you have in mind that like, where would be a good place for us to be from a browser in connectivity standpoint at the end of the year.

C

What is that some, like that total list of things you'd be like that's December I, feel good, because we got all these things. These are the things that we that we got done this year. I would love to be able to have that list, so I think right right now we have 13 minutes left. Can you write down some of those things in this in this item of my company? That would be in that list. What would what would it look like?

C

What would a good good year look like from from a connectivity standpoint from our browser integrations standpoint so that, as we go into things like q2 planning, we can refer back to this list and say do do any of these things we're listing as our quarterly okay ours? Are they getting us measurably closer to this, and even better would be to say, take this list of things and walk backwards, say what what are the things that would be need.

C

We need to do for this to actually be possible, but were these things and we could walk backward and then what gives you that gives you kind of a loose draft? Okay, our list for the next couple of quarters. That gets you it's at the end of the year. So this idea of really thinking you know like we're I feel like we're.

C

Moving from a place where some of the fires are now being tended they're, not as on fire anymore, we're getting to the point where we're releasing looking forward like deprecating ago stars, the thing there's a very close feels like we're. Soon we have you know like new new connectivity options coming up. What does it look like past that we can actually start to look forward instead of like just being in a place where we're like? Oh, we need. We knew, and we know all this stuff needs to change.

C

We just haven't got to it yet eventually, we're gonna be out of that position, and I want to start to shift us thinking into shooting a little bit farther ahead, we're where we are right now and looking at how we would get get to this that get to that point and a little bit more intentional and organized way. So a lot of times we get to qtu planning or next quarter. Okay, our planning like what is okay, what's that was this fire to fight, but instead I already have a list of things that were like.

C

These are the place things that we know we need to do this year and here those steps that we need to get there. So then we already have q2, q3 and q4. Oki is already pretty much pre-written for us. If we have an understanding where we want to be at the end of the year. Does that? Does that sound, crazy.

B

Typing typing machines.

C

Oh wait: this is this Eliska to get to log. Should I restrict you to 15 items per person.

C

Hey these look, like good things, I'd say, did feel free to stretch too I.

C

Mean like stop and stretch or also add stretch goals which are we just you know yawning yeah Andrew, it's too early ooh.

C

Browser notes: don't carry a big crypto bundle with them. The more I learn about that situation. All right double, maybe even triple encryption.

G

Rainbow, that's a really different feel.

G

It's extra security, don't worry about it.

E

What about they gave a scene.

C

That's something that really goes across project right or is that something it's really specific to browsers? I think it would be more of maybe a developer experience. Okay, we have. We have one of those people here, mr. dias Senor Diaz, so.

F

The question the.

C

Earlier conversation around API, versioning programmatic, API versioning as a developer, experience question because that's not a browser specific thing: that's about stability of the programmatic API across all the different deployed environments.

F

But in what you want to version the programmatic API, that's hard, we're.

C

Talking about it, mmm we're thinking about it, okay, well, would love to hear your thoughts.

A

Like the core interface interface repo, the idea is: how can we like expose the version of the package that already exists and make it useful for people who don't want to break anything? If that programmatic interface changes.

F

You can expose that version from the version command, yeah.

A

So like we already have protocol version and the daemon version we could at like the first yeah.

F

A

Jas ecosystem yeah.

C

Kind of takes during a slot machine where you pull it down and it spins through a protocol version demon version, the SPI version you can see where you land.

C

We have a lot of, we have a lot of configuration and meta configuration.

C

This came up in some conversations with other collaboration partners as well, where it was unclear out of all our configuration stuff. What was what was kind of did? What were the defaults and that came in out of the box.

C

Hash all the version numbers together, multi version, that's it.

C

You add multi in front of anything that goes a different meeting.

A

I'm pretty sure, that's an issue about multi version.

D

Just called meta version: it needs to be multi.

C

Alright how's this: let's look can I interest it in like how many of these are nice to haves for, as must and then separately, how many of these are like a stretch crazy for for the year versus, like oh yeah, yeah I already expect to get that done. Should.

D

We use random emojis.

A

C

Random emojis and.

G

That way, you don't know what the book is up to your interpretation of what's important,.

F

C

Development feels like an interpretive dance, sometimes I.

C

Like this p, zero stretch.

C

Yeah the DHT stuff I feel like we can decision on DHT in the browser is something that's gonna happen either way. Actually that's DHT in KS, not necessarily browser right well,.

B

Yeah so I think, like the DHT in J, ass is probably going to happen. I would wager that that's gonna be the case Ichi running in the web browser- maybe maybe not so much, but if we can reasonably keep the number of dials required for DHT query down, it becomes more feasible to run a DHT client in browser, so I think if we can target like okay.

B

What is the reasonably looking at query times like backtrack out of that, like, what's a reasonable query, time backtrack out of that I figure like what's the maximum number of dials that we could do to stay under that threshold and then see if it's even feasible for us to get the DHT queries under that threshold of connections, though I just put like 20 dials in there, but that is just a random number from my brain internet.

F

B

Be based on real numbers, I.

C

That that we do already kind of know it sounds like you already know what, like the bite-sized pieces, the questions that we would need to vote, a answer to see if it's a feasible thing, somebody that's in him having.

B

Bht notes that are actually diable from the browser.

B

That's another tricky problem.

G

We'll add it to the list there you go.

C

Great, this is an awesome start. Thanks. Y'all I really appreciate your your thinking on this and then we'll use this as a reference when we go into q2 planning, which is start pretty soon so start thinking about. What's what's coming next and and think about what are the pieces? That would lead us to these things as n states by the end of the year, and what we'll do is when we start to do.

C

The q2 planning is kind of look at what the sort of OPRS that people have drafted and they're proposing and then maybe even just like link them to this list and Lydell from a kind of browser and the connectivity organization standpoint. Maybe it's worth filing issues for each one of these if there aren't already and then which there probably are and then and then then making sure that we're tracking against those issues in each one of the old card lists, of which one of these that effort would support.

C

And eventually, then, we'll just have the hell out of it: jacob jake' style.

A

On my food list.

F

If an updating.

A

Readme of in web browser yeah, it will happen eventually we we should start the projects and probably as soon as I, catch myself from underneath yeah.

C

I'm gonna doing a kid to make that easier to like I said for the it was my goal, for you know that we had these special interest groups kind of spun up this quarter, but next quarter we're entering that quarter with these groups already kind of launched it running.

C

So I wanted to be able to have queue to really be the kind of line in the sand where we start using the same methods that you're using for heard for the protest ground and content, writing and things like that and have some uniformity and consistency from across the team and how we're tracking this type of work are.

A

We thinking like a separate, send, isn't about spare especially interest group or for a super.

C

Group really, because there's just like make a lot more sense, I doesn't think for that the whole ecosystem one would just be to high level and across the stupid these streams. Each one of these groups are pretty self contained in this set of work, even though sometimes they relate they can run pretty independently and I. Think the team needs the full view of just their stuff and then maybe.

F

They'll have a ecosystem.

C

One that's just towards all these. These okay are in the viewer, meta issues, or something like that.

A

Yeah I've run into similar problems with defining back in the time when we were using waffle instead of some board for sort of like high level view of web browser things. The moment you include like Jess, ipfs and Jess I, give a safe city clients on top of those like meta repost, it's instantly gets too noisy. So the special interest level sounds about right. Alrighty,.

C

C

G

What are the best hours of.

C

G

Next week,.