libp2p Team Meetings, 3 Feb 2020

Previous Meeting Next Meeting

⏯

youtube image

►

From YouTube: libp2p weekly sync - February 03, 2020

Description

https://github.com/libp2p/team-mgmt/issues/16

A

Alright, hello, everybody welcome to February and the lipid EPE weekly sink. Please add your notes, weekly sink notes. If you have an agenda item, add it there and we will prioritize that, otherwise we will go through the weekly updates I'm and since there are no agenda items yet, let's kick it off with our Sh.

A

You are muted, silent, updates.

B

Yeah on my side, I finished, creating the ps4 introspection 0.1 work. Router is now reviewing that I have also you know, finished reading all the documentation we had on content resolution to get up to speed with the plan and the issues. My focus now is going to be some lip v2p bug fixes that my can now have asked me to work on youssef and also address the introspection review. As now starts posting, it.

A

Sweet any questions, harsh.

B

So one thing to add around I think your mission in the chat today there are a couple of comments: I left on the core PR after you address the good or commit just take a look at the yeah.

C

I saw those yeah yeah, awesome, I saw us, go Thanks.

A

Okay, we did have an agenda item pop in so we will redirect to that. So we make sure we have time to talk about that a Dean. Would you like to talk about your question? Yeah.

D

So, given some of the upcoming changes to how the DHT is working, there's this fine pure function, which ostensibly helps you find a peer so that you can connect to them.

D

How this function should work I think will make change as a result of these other things that are going on right now, it's just like as soon as you hear about. As soon as you hear about a provider, a super hear about an ID and an address for them, you're like okay, I'm done, and then you go home, but what if the address has been updated so there? If you're interested in that question and more then maybe well I, think having an having a discussion on zoom instead of github. Things will move this along faster.

E

And where is this discussion here?.

D

Now I I don't know not now, because we got the other things to do, but I figure we can set up meeting sometime this week. If people are our game, yes, yeah.

E

A

You are interested in attending that add your name under the agenda item there, and then we can coordinate a time for that a Dean. Do you want to coordinate a time for that with everybody.

A

Alright, are there any questions around that.

A

No all right, we will jump back into the async updates.

A

Next up is me yeah, so we released the J slippy to be async refactor last week through that over the wall need to finish up a block post for that this week, I'm also doing some profiling on the release to make sure everything is good to go this week, I'm going to finish up that profiling and then start work on integrating the J slippy DB noise, repo that node factory worked on I'm, going to start working on integration with that n2j slippity P and then also working on jslabbee to be twenty20 high-level roadmap objectives so should publish something that this week for everybody publicly see any questions for me.

B

Jacob, when do we had the next meeting for content resolution estimation, Booker I.

A

Will look at scheduling something with later I? Don't want to start that too soon. Just because like having two meetings back to back is are too close. Together is not going to allow us to get work done so I want to make sure that we're we're not blocking ourselves on work.

A

All right, a vascular, everyone.

F

So I got back from holiday today.

F

Basically, today, I got through my email and I think we Jacob about some of the pyramid, roadmap and objectives for this part and narrow, so more long-term, stuff 4020, and for this week I want to basically get some work that I got post when I went on holiday. I want to get the lipid to be. A key chain goes right into with jessep HP I already revised the PR and basically Jacob will review it, and then we will merge it.

F

This week also I started working on getting the interrupt tests running ingest the pitch before each peer and basically it's the same thing good revised and hopefully get March this week and finally, I will continue with the Star Destroyer factor in order to try to get it ready for review during the next week, and that's it for me any questions.

A

Awesome next up, Yousef, okay,.

G

I made more progress on peer records. They're very close I missed an important piece of feedback from rule about the API I'm gonna finish that today and then there's pure story changes that are just another loose thread, basically like I got to red, green and the red green refactor cycle, but it's all messy and way more complicated than these two visa. That's another hour so today to clean that up and then hopefully we're thinking tomorrow.

G

Maybe Wednesday of things merge in the Pierce gear record changes so this week for sure, then, noise stuff is basically where my my head is that otherwise I did some minor things. I am actually got some numbers out of it. It looks to be a bit slower than TLS, which I think.

G

Probably the majority of that is explainable by my computer. Having hardware aes support and TLS is using AES for its block, cipher and reason ChaCha poly, which is probably not as fast but that's you know, there's also lots of other stuff that I kind of want to fix up in the noise code base, because it's kind of messy and hard to follow and reason about in places and I think there's some improvements we made there.

G

So I might try to thread those in this week, alongside other like more like quantifiable goals, which is just well at first there's like quantifying exactly what needs to be done before we're gonna, say: okay, we're pretty confident about using noise and production.

G

So I have some issues kind of loosely rolled up into milestones, but I wanted to spend some time actually reviewing that and making sure that everything is covered and that we have like a good plan that we can kind of eyeball how long it's going to take so that once I have that all written up. You know we'll work through it and try to get like a release candidate or like at actual something. We can tag as it released for noise and start.

G

You know using as a dependency elsewhere and then also bug fixing with Irish we're going to try and coordinate on that there's some bugs that have been open for a minute, but you want to yeah.

A

Great any questions for Youssef.

F

H

Any more changes: the API because I merged the P R in the PR for the P records. Okay,.

G

If there are I, I will definitely check your branch make sure if it's affected I, don't think that it is because you're not creating any peer records in your branch and the only change method is the one that takes an unsigned record and wraps it in an envelope and signs it. So since you're, just like verifying and taking the record for like the local peer record and sending it on the wire I think it should be fine, oh I'll, make sure before I merged.

H

Okay, that's cool, then. Perhaps when the the Sentry records old nerds, can you make in another PR that actually points to the right? Go mods and everything definitely ready. Yeah.

G

H

I

One to be usable, yeah, okay, great all,.

H

A

Sweet any other questions for Youssef nope, all right, Mike.

J

Sorry, but that couldn't yeah this week, working with the arsh and Youssef on the security bug fixes. So just these are we don't know of any exploitable.

J

Vulnerabilities or anything and Lippi to be, but these are some long-standing things that we want to fix and in pretty good what we want to publish a transparency report about those that have been reported and in order to do that, we sorta need to close loop even on things that we don't believe, are any significant. It's the real world, and then this this.

J

Some of this is like this or next week, some revisions to the it okay, ours, based on some new things that came out last week and then I'm hoping to get finally to UM what I thought I would be jumping into right when I came back, which is trying to build up, build up our team and on board more people.

J

Yeah and yeah: that's basically, it.

A

Alright, any questions for Mike all right next up roll.

C

Hi everybody again so last week, most of my focus was on test round on landing compositions, but I did have a bunch of interviews on a bunch of meaning across the board. So that's landed now, which is great news. The this week, I'm gonna be focusing or actually just like tomorrow, mostly I'm gonna, be focusing on closing off the whole run.

C

Events, outputs test plan instance how couldst assets artifacts affair, so there's a unified model because they've been like a lot of different directions that we've been taking across the the codebase, and this is created a bit of like spaghetti code, so there's a unified model now and I'm working to basically create like really solid API.

C

For for that last week, I also attended the e 2.0 networking hall, which was great once again, I'm gonna actually paste a link to the update which I forgot to do cuz I, send them a Anna synchronous, update and just ran her it I'll post it in a second for you guys to to to benefit from that as well, then I, we closed off, milestone 0.1 of test ground this week, we're gonna start planning Mars from zero point two with Fabio who's, a new person from Moxie. That's gonna be working with us.

C

That's the technical project manager for for a test ground, so we're gonna start that they're, probably going to adopt like a sprint model. Scrum based model we'll start experimenting with that, because we've had a lot of success with the two-week cadence. So what probably would probably move in that direction?

C

Also reviewed signed peer records, as you have said, there's a few things to to close off there, but it looks really good and I'm. Pretty sure that you know you have like finishing. This off will probably end up landing it tomorrow, day after tomorrow. If all goes well, so that's a great building block for the future.

C

Then discuss content routing one aspect that we identified as a preliminary dependency and the entire plan for content routing which was getting network forming for test ground.

C

Well-Designed worked with that with what Stephen and prep for the security audit and surprise surprise started testing before we ran this, so ours created a script to automate the the issue transfer from children. Purpose of gonna be to be to the top level wrap up and before we had rules in our forest of issues. I decided to like do some model like some real world real world circumstances that we find in issues.

C

So this is PR is that link to issues commit that links to bending to issues, issues that intimate with one another and like this like complex graph of dependencies and links and decided to do some manual testing by transferring issues manually between repose and github, just fell flat on its face. So basically, we have two issues that are now 500 Inge, all the time, so I hope. In the case, it's not great. So there's gotta like puts a stick in the wheels of like overhaul.

C

You know project project management, revamp planning for a levy to be because, like we can't have not and not immediately a you know, backlog that we can reason with for now. So it kind of sucks a little bit I'm hoping that if somebody has a contact here at github that you know you can you can pick up pick up. The phone call pick up the phone call and kind of like have them look at this. That would be amazing.

C

Another thing is that I worked with 0x tool and WebSocket secure dialing on the goal it on the go to b2b transport, for that when I'm, not sure we'll need this in jas, especially if we're shifting focus to the browser, because I mean the WebSockets api's are what they are in the browser, so you can't really like fiddle with them. A lot I will need to look into this again for this work.

C

Besides, what I just said, landing a unified motor for test outputs, onboarding, Fabio planning tasks, remonstrance you're, going to I'll be doing more reviews, reviews that are halfway there and reviews that you just start and I'll pick up. Some content body work so for me.

A

Tweak any quest for Oh.

A

No, all right, we are at the end of our update list visa, or do you want to give an update, yeah.

H

Working on procuring holes, design document would spend you know like working on figuring out how to test a scoring function. So so far, I have three specific attacks and I'm going to test with simulation. So there are 56 scenarios. So the first scenario is general squirting attack where noticeable nodes just appear in the network and do nothing. So we need to see how well the supporting funds react to that. You know they're, basically, just what in the network.

H

The second scenario is clipped attacks against a particular node, so if Sybil's appear and try to that particular node by performing in graph data, so one see how the scoring function we actually- and the third scenario- is more gross. It's the Eclipse attack against the entire network, so basic the attacker creates a whole bunch of Sybil, denotes the Sybil note from Graf attacks against the entire network, so over the state.

H

How, well you know like the scoring function performed there, so so the objective of the deaths and simulations is basically, if you note what a good parameters, how much time it takes to recover from this kind of attacks and firm well and generally, how robust they did so I'm gonna put those into the Google Doc I have them in there and Emmitt's buffer right now, but like I'm, open for suggestions for additional tax might wanna simulate.

C

Sounds like a great start, yep.

E

Any questions comments for visa yeah so either do you want to start migrating any of that into the overleaf document that we had? Oh, no just yet I'm.

H

Gonna put it in the first scoring document, so basically the score function, document yeah.

E

H

So collected there, okay and then by using figures or at least parameters with regard to the missiles rates, because you can't the parameters can be relative to the missiles rate. So, if you can't, you know, like a high emissions rate network, will contain one value for the parameters for low missile straits. In the network you make, you know like a different value. The parameters yeah.

E

Sure so you should I just start getting that into the more paper format version that we have. Oh.

H

Yeah the paper about specific attacks I want to address, do that yeah.

E

Okay, okay, yeah and, regarding the previous version of the description that we had in the gossip table, like technical report, are we ok anything now to start sharing that with community or put it on the website or anything similar?

E

E

Yet I was just yeah, don't read.

H

All right, when I do a small update for the specific attacks I just outlined, so you know we actually list the mouse possible attacks for here considered. So perhaps a community can come up with new attacks exactly.

E

Yeah, so my thinking was not necessarily to put them the scoring function itself funny until you know we decided on the final design, just the general description of what attacks. We want a system to be able to yeah.

H

Well, we're definitely we definitely want this into the paper. Okay,.

E

Okay, okay, so, okay, okay, fine, okay,.

A

Sweet did anybody else want to give an update that is not listed.

E

Yeah, just quick one for me not exactly an update of a little bit more on them outcomes. From the workshop, you should have received a couple of emails for me whenever those in the workshop.

E

More than welcome to give any feedback flagging issues so that we can move that in parallel.

E

That's one thing and the second thing is there was this issue on DHT robust Clips, robust the HD, and there was a discussion there I started reading through and then started doing a you know, related work, search on what exists out there.

E

Should we? How important is this issue? Should we spend some time to try and okay.

G

K

Just an idea that I want to write down, it's not.

G

Something you're tackling right.

K

Now, like we can, what could do I know? There's we did get some feedback I Becca, chair separately, bye yeah. It was three I'm I do yeah.

E

A

Any questions comments for us, no all right if there are no other updates once twice sold. There is one other agenda item from Raul.

C

Yeah so I mean it is almost gone unnoticed that the jlpt released a single wait. Thank you guys.

C

A

D

A

C

C

Please unmute yourself and I give a huge round of applause, because this has been in the making for a long time and we finally shifted. So please a mute and give a big round of applause to the J's that the B team for shipping a sink away.

C

A

Yeah we'll be pushing a more public public notification, but we just wanted to. We had all the bits and pieces together, so we wanted to get it out the door for all the early consumers, but we'll be doing a larger, larger push. I, don't know boss, the singing but yeah.

A

All right that is, it did anybody have anything else they wanted to cover talk about I.

D

Had a question about condemn Leah bootstrapping I'm wondering if the function as it is written currently was designed to work with the DHT as it was, or the DHT. As the kidnap s condemn Leah specifies.

C

Can you unpack that a little bit, what exactly? What exactly needs you.

D

To um yeah so I'm not sure it walks towards itself and then afterwards goes and walks towards all of the other buckets and I think it goes and sort of decreasing orders of closeness.

D

This answer that seems like the wrong.

K

Ways I think that's the right way so like basically we're trying to make sure each book is fully so.

D

We do a very shocking.

B

We start by searching.

G

For self, because I find out how many buckets we see.

K

D

I'm just wondering, if, like in the boot strickler, depending on how many nodes you're bootstrapping from I, feel like you may end up, you may end up actually wanting to call it bootstrap on self more than once, if you're like, starting with like one node in.

K

Theory you shouldn't in theory, if you call it once it should go to the end and find different supposed to you. If it doesn't, then there's a book now I'm, not sure how we do this. We may have a bug here. We're like we should probably look at like how close four closest beers like so instead of looking at real buckets like you're, sorry, you didn't look at the buckets we have, which is like he's like combined buckets.

K

We may want to put it like so logical, buckets yeah that is likely assuming like we have like one bucket that shares one bit one, but if there's two bits well, etc, etc. Think that was fixed, I think yeah.

D

Okay, yeah just just check in I'll. Let you know if I find anything more concrete.

A

Sweet we are almost at time. Does anybody have anything else? Last chance.

A

A

Turk has notes on your own idea and refresh buckets.

A

All right is it. We are just that time. If you have anything else, it's too late go ahead and hit up github. Alright, we will see you all next week and bye.