IPFS Project Operations Working Group, 22 Jul 2019

Previous Meeting Next Meeting

⏯

youtube image

►

From YouTube: Project Operations 2019-07-22

Description

Weekly call to discuss project operations work.

Notes: https://docs.google.com/document/d/1IZ8tyWKnuXKTMbYaZO_8FVUKJ1kxA4imgPaTBhXXbAc/edit#heading=h.pozptkwiy7pi

A

Okay, I guess we can get started so initiatives. First, oh yeah principle: agenda. Okay initiatives upgrade to go up first, release process. This is me. We have entered into stage two release process for I, go fs0 much like we already have early testers running it in deaf deployments. We will probably enter stage 3 next week. Assuming nothing goes weird great. The testing process deafness also may then Russell is missing team.

A

We now have regular meetings. Every Wednesday, 18 PST here at the to local timezone Allen, is currently reviving our best benchmarks and then we're gonna try getting go testing as well. This is currently this is built JavaScript, but they can work for both a role that has done some architecture. Design work for the test, lab, I and Jim has done a bunch of work I for like there's like UI mock-ups.

A

We still need to like make sure we sync mean, like your we're, gonna, pull it out of that meeting and sync up I'm like what this benchmarks currently does, because we can probably reuse some infrastructure. Some of the UI I, depending on like the tests, were trying to run I'll hand it off to whoever wants it about Jim, pitting adjacent. Yes,.

A

This might have been an ally doing.

B

C

It is a Dirk yeah, I can talk about that, and so basically we're trying to just kind of close out that garbage collection, PR.

C

That's it's been running for quite a while now and so with Alan I was talking about it and we thought it would probably be best to try to get something something out there, because it was kind of like a lot of ideas about improvements that can be made to it and ways of doing it better and so we're just trying to kind of narrow down the scope to a minimum, labeled garbage collection and then open issues for her outstanding stuff and I'm kind of splitting.

C

My time between that and wrapping up on improving might be a bad performance for your idea. Bus.

A

For delegated routing is that a allen thing and allen happens not be with us, is that's both an Ellen and I.

B

Can talk about it.

D

Yeah I can hear me yes yep, so it's something I've been looking at for for my integration with brave. Basically, it's a nice stopgap.

D

Until we implement actual beautiful transport, we would ship embedded, Jess ipfs does delegated routing and I've been debugging performance bit deeper last week and today and basically we need to like that, there's a performance problem of how those modules are implemented right now there are no trolling and we have the limitation of how many concurrent requests we can do for a single domain in web browser and also we are doing swarm, connect constantly, which is not necessary. At least we could shuttle it over some time windows. So I got some changes.

D

Luckily, that I will be our that should address that. That's a separate topic of gateway, API commands configuration option in go ipfs, not wired into anything. I talked about this briefly with Steven, but like it's like, eventually to be worried up, but for me that immediate need its to get those little to be modules responsible for delegated routing, to the point where we can actually shoot them with Jessica. First, the current version will just float not only the web browser, but also our preload nodes, toggle DDoS protection.

D

So when I decrease that the flawed it still got blocked on the other end, so we need to. We need to figure the out box library and also configuration of our pre handle a lot of requests, because we can, we will decrease the number of requests sent from individual node, but if we deploy it as a default or like as opt-in and people by default will use our field notes, just like they did with WebSockets. Tar will quickly run into similar problems. I guess: that's like current state.

B

B

Think rush was out, but he's working on I believe getting the fallback implementation finished up the base, gossip supplementation is done and I think we've got the work done for falling back to flood sub, but we need to finish up interrupts tests in forego nsj and then we should hopefully be able to release that soon.

A

So I guess it's up to me. It was known that would be Logan.

D

Yep I believe we sort of got a better idea how to handle this. For the local host had a sink history Stephen. We there's like a sort of clear, clear path forward so for local host isolation between subdomains and path based gateway. An unknown problem is how to handle those all those bash scripts that expect half base gateway, I mean if we suddenly start redirecting path based gateway and point to the subdomain one that will introduce HTTP redirect and a lot of scripts out there.

D

The wild are not expecting redirect and will just crash, but that's something we can mitigate by, enabling redirect only for web browser user agents. Initially mostly will like will care about origin separation in the web browsers. So that makes sense.

D

It's anything. I missed, Stephen, I.

A

Well, this one will think you're a variant encoding, parodies series I'd like to get a look at this. Of course, is people really put this recipe folks? So I guess is here even so: no news, okay, second coupe or here I yeah, basically to office. We can either distantly to be and make this work for all say IDs or we can do this or sorry all PR ideas, or we can do this as an hack at the gateway I'd like to do it down lower just because makes it simpler.

A

So I guess next is replicas validation,.

B

Yeah so similar to what go, let me repeat: core did I mean holiday at all those repos we're going to do something similar with J slippy, to be core modules and then with the J's will be to be. Interfaces will join this someone to be working on that this week start with the interfaces, because that's the least obtrusive thing and then work my way to KS and that'll. Probably just do a repo consolidation first, but then look at what we can do to optimize tests. Steven I.

A

Have a good concern here in yo ipfs go core. Are the interface.

E

A

I want to make sure that we don't have situation where, like that's our interfaces and then GS has an inch base of rapid core meet something totally different confusing. um So what do you mean by court? You mean like core primitive support like what so.

B

It will be so it's basically all of the modules that Jas lovely to be needs to function by itself. So add the connection manager repo, not the interfaces themselves, but the actual code. We're gonna, bring this new one central repo. It's.

A

There any way we could change this around I'm just like yes, this makes sense, but it's going to really cause a lot of confusion because it's a go. We have all benefits in quark who say like this is they're all interfaces and our like. Basically instructors like PID definitions, all that kind of stuff in this one repo. So it's like the idea is that each building or import any liquid, P implementation, but they use the interfaces in court. It's like a standard thing just goes like are.

B

You just getting hung up like. Is it the naming that you're you getting Stern about it cause? You know it'll just be J slippy to be so. Will we pulling those sub modules into the libertage, a slippy 2p repo.

A

Oh okay, okay, servers like if you have a DSS for repo and it definitely very good, go then go to the free desk or you know.

B

So well how it will end up with JSL p2p, which will have those other those space, those core modules in it, and then we'll have the interfaces repo, which I'll have all of our our combined interfaces. Okay, wonderful.

A

Then I guess for now on two blockers and asks I have one asked I said: I have a lot of one-on-one 7 Mondays to distribute knowledge and have time you have to stick with people so that I keep this on my plate, invisible! If you want one thing, we I guess next, one Hannah is wondering if there is a JSP or person. Okay, they do exist.

A

Who is that I guess.

C

That's me: I'm the repo maintaining.

F

Tim week, who is like I've just been brought on to do Jess bit swap I, don't know they're still, there did I said hey you.

C

F

Up, oh well, honestly, I just want to make sure you if you need, if you ever need time to check in with me about like things that have happened in your looking to bring over to Jaso ipfs bit, swap just I just want to. Let just say that I'm available. But you should.

C

A

Any other blockers or initiatives.

E

Now I even talked briefly about where I castings are at right now, which is just there is in my me list down below there's a PR, that's tracking, all the pub/sub aquinnah stuff. There are an issue tracking all the PRS, there's four that are outstanding that are like. Basically there I showed a demo during drying. Ipfs camp just didn't get those over the line, and then we probably will need to start working on like these features of IPs that we have neglected because it was slow like local pending.

E

There have been some pre-existing issues about that I liked, one of them that just you should be able to start up your note. In the same way you have penned content.

E

You should have ten by pianist records, it's as simple as that, and then there's there's sort of two tangential issues like one is: maybe splitting off IPS from I give us, because there have been requests from a number of parties about that, just being able to use that separately and then also figuring out like how much and when do we need Jas, go compatibility between these things.

E

That's the current state. There's any questions.

A

A

That's it unless watching a parking lot, I.

A

Because I think weighs about that. What.

F

A

He wants to put the pertinent issue.

A

Many years coming.

F

F

Now we just need to see it fail, what production, without going I mean or like not fail like.

C

F

A

Okay, I think that's it. I have a wonderful week, see.