IPFS Weekly Call, 26 Dec 2016

Previous Meeting Next Meeting

⏯

youtube image

►

From YouTube: IPLD Sprint Call

Description

This was the IPLD call on Friday, December 23. See https://github.com/ipfs/pm/issues/293.

A

Recording now yeah.

B

I think I think we lost Johnny I were talking, I think he disappeared. Okay,.

A

Well, I know he's just listening and welcome to I feel D. This is like our weekly update, your probably things on YouTube. Later it's no one section streaming. This welcome nikolaj greco and me. So we have a few things today we had joining crunch on the call earlier. I don't know if you dropped out or not, but I guess, let's just take it away with the notes and if Johnny goes like on will go to his point, yeah, so stay website, yeah.

B

Last week we mentioned that, and there were a few things left to do and I wonder what is this I've been following, but I wonder what is the state of that right now and how comfortable with because it's just me and you let's go through the things that only we're about me and you for a second and so I read your rewriting and they wonder if you read my comment on I did.

A

I actually, first thing this morning and uploaded a finished version of my comments. It's not done. I mean the sections that you have to do, or it's still there for you to do. I didn't keep writing you want. I can or I can try, but for now, I've basically gone through I. Also responding to your comments. Okay,.

B

No I I see the sponsor you my comments. I, don't know if your. If you feel that you're done with that or not yeah.

A

I'm doing this as it's now your turn to keep writing more than I. Could anymore now, I feel: okay, okay,.

B

Because I want to make sure that whenever I jump on the plane, I have everything I need because very likely. I'm not gonna, have internet connection.

A

Beside it to you, I decided to me I'm done as far as I know. Okay,.

B

That's brilliant because I remember the for the web said we were missing very few things. One was the demo from Victor one was the content plus the example from me and I think that was it and some beautification steps but I? Unfortunately, Victor is out and I don't blame him, because it's the day before the day before Christmas and I.

A

About to say so,.

B

I did I'm done, yeah yeah I, even I, haven't had much from Victor, so helping him I be viewed as I mentioned last week. It would be amazing if we have something to show I get feedback feedback from the lab week.

B

Okay, so I, let's move the next point later on. So serious back of paper, he's gonna happen before the 7th of January, because that's our deadline and I wish I think the purpose of today, in my opinion, is the okay. So the tutorial will discuss that. Okay, all things were very easy.

B

This is just like checking, especially because every flight bad but I, want you to look about next next year and how can we improve what we've done this year and they will work this year from next year and I wanted to sort of build up. I have a wish list of things that we should bring up a lot week or in the future.

B

B

We can bristle together. I have some clear things that I want to bring up at the lab work and amongst those I want more presence from from the wit, because is implementing all of these things and he's going on the speed of light and I'm gonna be more connected. Maxima nest next semester and by need to make sure that I have at least a weekly meeting in which the wave is in.

B

Meetings is there something else that we want to raise, maybe having more clear deadlines.

B

Having someone that leads with more time, because I think it was a bad idea to have me having lean in this with very real time, we can do I, think we kind of made it but I think we need someone that makes us move much faster if I'm, if I'm gonna be seen as next semester, I don't know who that would be, but that's something it's critical to the project.

B

Is there something else you miss just in terms of the way we work, then I want to also what you should work on Mac semester next remaster next quarter. Sorry, is something something else that you won't raise about: I, don't.

A

Know I mean that sounds like it's me registering get it out talk about it. Talk about this behind me. Having a conference a little plan. This plan website should be launched, yeah almost working the most that's it.

B

Okay, in terms of.

B

We say what I want to make sure that we have some some stuff to raise in whatever is gonna be and for next time. Next there are some particular things. I wanted to look into and I think many people are talking about this I peel, the selector and they're, raising very good questions, and especially the the guy from a medium is called cumin.

B

Yeah, yeah and I think that, amongst the priorities of the next quarter, should that he'll be selected and and what that entails, is it's not just the IPL be selected by it's sort of basis for transformations, because we've been talking for a long while, but we always put that on the side, because we wanted to shift IPL be first, but I think it's now time to try to achieve something, especially the IPL dislike. Having that be L dislike to really make a lot of things better, especially for the Athenian people, as I'm saying I,.

B

Really want to have debate in these meetings and not just a bit but possibly sometimes one being present, because, especially for the website I remember last week, Victor was a bit scared, the one who said to take a different direction and he wants to plan a meeting and we didn't have a meeting and we kind of lost one week anything okay.

B

This was for me just check-in. Victor is not around so I, don't know what we should I.

B

Want to check with him how how I felt about the work that he's done, and he feels that we had never. There.

A

Hello I think he's waiting for once feedback. There's some things I think we're waiting for you to have more time. I'm just here to copy other yeah.

B

And also you have all of us thinking I think this very important way I would not have had this meeting myself. So what did John you mentioned so that we can still that we can think about it. So I.

A

Posted an issue SP asking who use cases and demos which we need to have for the website. Johnny crunch, actually posted pretty awesome, late about exploring, doesn't know because also visual identity solutions.

A

There would be great to have months fall for you there, because I.

B

A

B

A

Timing, I, don't know how you did that. Can you describe to Nicola what your demo could be or is.

C

This for me, Johnny yeah,.

B

C

I think so sorry, yet I'm at my folks, house and I think it's just both my kids are streaming Netflix at the same time, so to get back on so uh so yeah I think it's so within identity management, I think be, there's a need for right now with json-ld. You actually have the context. The context needs to be a persistent URL. It's perfectly application for ipfs or IP LD to actually to link the data for a permanent source of the context.

C

So if you want I can actually just me just the tunnel back into my other machine at home. Give me a second and I can sort of show you an example. So.

C

All right, let's York.

C

There we go so.

A

Is that the Jolly Roger.

C

Okay, so I, guess you see this mm-hmm so yeah. This is a typical consensus. You port token using Java web tokens with signed tokens, and so this is both I think a two tokens one is a public key and the other one is actually a claim. So here the payload is claimed with the name of Johnny crunch and so I think.

C

The idea is that, rather than using just loose attributes is I really sort of provide a context of all right where this for to provide semantic interoperability, so that attribute could be linked to other attributes. So in this case this is pulling from the schema arg4 and for a name and I. Think I can probably show you that so that's the I can get to it.

C

Yeah here we go.

C

So same thing, we're.

A

Only seeing the single oh no got it.

C

So yeah, so here is the context so which is a JSON and it provides in this case, identity, attributes and linking it to the the schema of person schema. And if you go scroll down there's this is the name so schema name so, but it also could be a matching to first name or a given name, and so the the the challenge would be in and let's see, if I have.

C

Word yeah, so here's that same thing, but within the claim it really would be two.

C

And json-ld would be to provide context, and that would be the ipfs hash I think. Actually they would do and so I think that's. The first thing opportunity is actually providing like the context linking to the.

C

Schema and then the next thing would actually do be to provide an address and so I think with IP LD. You guys are working on linking to a particular block chain, that's stored on ipfs and so I think the other nice thing would be to actually to have the in this case a you port ID, which is really an account on ethereum, really sort of points to an account on any particular.

C

Blockchain store night PFS, so my.

B

Question is what, if this cannot be already done, I think I think what we're describing yeah I think I'm, not sure if you can put in additional be a context inside the sub object, but let let's say that you can, then you can already do the app context in the way you did it right and and instead of using the Newport hash, you should use an an IPM multi hash and that would probably make we want or doesn't listen. This is help.

C

Yeah so I think I think right now be the multiplied hash for an account. Do you guys actually have that from multiple block chains, so Abel honey resolved that so.

B

And now I, don't know exactly why you pause I'd be like I guess is the hash of something yeah.

C

This is a theory on account in this case is actually like the chest just yet: okay, I.

B

Think there is some more going on I'm, actually I'm sure there is some work going on with this okay in IPL D. We have this concept, which we haven't fully explained on the website yet, but this discount of CID is the content identifier and in this constant content, content hash world content vegetable world.

B

We won't see ad to be able to content address anything so, for example, so there is a way in IPL, D, 2, so I see these basically a hash prefixed by what type of hashes this so I said there are some CDs for the idiom blocks and I'm pretty sure that you can reuse some of that work to say this hash, the hash they're pointing to you can prefix it with a the type of the hash that the hash that the hash table is and then you using instead of writing the ash in that particular way, use multi hash.

B

So we know if the hash is base64, we know it's the hashes. Why is the hash function, and so I think by replacing all these hashes would see at this would and would say this that we supposed then they would make everything work in with ipfs. So your concern now is first I want to put context in my objects, anything that you can do today. The second concern is I want to make sure that all the hashes that I'm using still work in ipfs and you can make them more by turning them into civs.

C

Okay, the one challenge is the CI DS is for the the table lookup, for which blockchain you might be on. So how, in the see IDs? How do you resolve this would be an address on the test net for aetherium versus the the the full nerium? Okay, okay, so or peanut, for instance? Actually so a private network so.

B

Is that you post I did the hash of a block of a publicly and address.

C

Account theory.

B

How do you create an account by generating some keys or by adding something to the blockchain, so.

C

You generate some keys: okay,.

B

Then this account is valid for a cross between Testament and not Testament, right. Okay, then, then, in this particular way, in this particular case it doesn't matter what blockchain there is, as if you prefix it with the etherium account resolver.

B

Then, whenever you ask for the hash to other people, and you can be solved, you can get the hash somehow from listening to a video or listening to a PhD like on FS, then you get that particular.

B

Your CID is going to specify what type of what I what this hash is pointing to. So you can say this hash points to an aetherium account and whenever you solve the hash, you read the content as if it was a theorem account and if it is anything account, you're gonna be able to parse it and you're gonna pass it as an IPL. The object, which is a good thing.

C

I suppose do you think you show me an example: if I actually would do that for CID with either etherium or, let's say a Z cache, so this is: how do you differentiate within CID a silly reimburses, Z cache? What.

B

I'm talking about is actually working progress so I have. mmm Is that mean I can point you to some one day it's happening, but it's really like happening right now. So give me one second and two: fine, it's not so not well documented, which.

B

C

Think, as an aside, actually one other thing actually worse talk to you right now within the Jason LD is nested signatures and I. Think with the consensus, the you poor protocol is nice, because actually these are just tokens and actually each token is is signed and attesting to the the claim is adding a new token and mostly it's actually using the subject and the issuer as the the Association type, and actually it's in the contract itself that actually that links the pointers.

B

Okay, for some reason, I cannot find. Okay. I cannot find the resolve that, for the accounts, I'm sure there is one guy which is called Kumari's. There is working on that it's in particular for ithi diem.

B

So this is give me just one second, okay: let's do this I'm gonna Richard escape damage chip and what I'm gonna do I'm going to post the links whenever I find them in that issue, like oh yeah,.

D

B

Did you add a same passenger me yeah.

C

I saw I think the other thing was actually were. One thing we're discussing within the verifiable claims task force is the nesting of signatures and the and I think the because every time you wrap a signature with a JSON web token and I think it could get to be very, very large and so I think it's just so I think that's one I think your actual we're talking about it's actually how to handle that and I think within consensus, I I believe, is actually pretty its handle pretty with tokenizing each of those.

C

So each claim is with each issuer and subject is his rap as a token and each of those tokens is that and then signed and it's attached to, let's say in the theory, name account or aetherium smart contract, and so the would been the json-ld world.

C

That's a little bit harder because you don't have the association, so I think the and I know that's. Why I, like the you ports solution to this I, actually met with them last week of a New, York, City and I think that their challenge them with is actually how do you make this scalable and generalizable, but actually it's more of a standard for identity on the blockchain.

B

Okay, I'm not sure exactly what we I get a gating or and and and and so what you would like to replace the. What would you like to replace with that token or miss that I.

C

Think well, in the case of nailed a world right now, it's actually you wrap tokens and so each almost like a PGP key I guess you actually like attest to the validity of the PHP yeah.

D

C

So I think the with what's nice about the Yuka for protocol as a token eyes it with a the attributes of the the issuer and a subject, the subject: it's actually a an identity on a blockchain. If you guys she can resolve.

B

C

Let's say one person is on a D cache and actually wants to attest, even though it's cozy caches the new smart contracts, you want to test someone's claim and that is actually on a different blockchain, so you actually build a resolve which blockchain is actually on and I think. Certainly a multi address, or a multi hashes, actually I think the way to go and so I think.

D

C

Send me the link actually gonna just hand how to do that, and so my point is that I think it needs to be a scalable and generalizable across from block chains. No.

B

Sir III think the links I'm gonna post, they're gonna I'm gonna have their daddies that we want with IPL D or in NCAA, be we want to be able to point to any hash. So right as soon as you have accidents, you can point and we solve any hash. Then it doesn't matter if this is on a blockchain on another blockchain.

B

Things of this kind, so that the aim is that we want to have this huge metal forest and that this Peaks all the same language and can address to each other in the same language. And then, of course, you can build something like.

C

Similar to so it's a similar to the mobile address for the multi hash. Actually, the.

A

Sensors do and.

C

So I think the challenge is that I suppose like so how many?

C

How many different types can you actually in a byte? And so, if you actually it's the first byte of the hash yeah, so eventually will you have a space in that first bite to actually the store or let's say there right now there are the block chains. So is there so.

B

This is a good question because it's a lot of arguing about this in the past. So right now, if you, if you won by it, which is 8 B, 8 bits, it's gonna be able to only support, 156, different types and then I think someone mentioned to use 2 bytes, because I've using 2 bytes would give us twice the amount I. Don't I, don't know what is the III don't know? What is the not res amounts are much more anyway.

B

I know it's gonna, be I. Think you're right your concern. Your concern is a same concern that many people have and also to give an example. How can new block chains be added? As with all these protocols are self-describing, so you have a you prepend and a back and up I'd.

B

It says what type that thing is and who right now at, if, as as like mapping from bytes to buy it from by to program and right now, whoever comes to us and say, say: hey I would like to reserve this by I can reserve it because we have this bot chain and so on. Then then we give Iran with with I, think the kindest strategy and see if we, if he's gonna, work well, I think at some point you will have to say not to some chains, but then they will have.

B

That would be totally fine because they can create they sort of say this and say they have like a sort of a private chain or private system. They need a particular and a particular type which doesn't listen. Then that's totally fine, because all these protocols are meant to also work locally. When you have your own table, but whenever we want this global table, then this becomes the tricky I. Don't think you solved it. This is not a question right.

C

Yeah, that's what sort of concerns me, especially if we're going to use like a peanut. So actually it had you resolve and keep the local table of a network to resolve across. You know the private network and so I think the you know, one solution, no in late in JWT, is actually separated with a dot and so and I think it. Actually. This reminds me of the early days.

C

You know guys you guys younger than I am but I remember the early days of the Internet of the nineteen nineties, so really very much the land grab of DNS, of the necessity of resolving addresses and as I think this is sort of where we're at with blockchain right.

C

That was actually the Sesa T to resolve and there's a lot of different camps that are actually coming up to actually to be at and so I think that may be an important to solve, with IP LD that if we solve that and that's the de facto standard yeah.

B

I think the the type of solution that, with by the way I just did the gut question. What were you were talking and I? Think just adding one more bite would give us 65,000 possibilities so I.

B

Of course we will add one more bite in the system of everyone which is a benign, but but this may this gives us the possibility to give the first the first set of by it's only for ipfs, approved things and then the other certify that they are by or the combination abode for more complex, so I don't think we're gonna run out of types.

B

A

B

Is how do we add new types whenever someone has a new blockchain that they would like that one? It's.

C

Possible I know the multi hash is the first, but is there a possibility to actually reverse it and actually have it as the tail and actually separate with the separation you'd like it and at simple another thing.

B

There are some strong reasons why we want to everything, because you know ahead of time before during the input, and so some.

B

If you, if you, if you, google, my content and I, add some question a long time ago and Kwan replied to me, saying that we're not we should prepare and I was really convinced that those reasons right.

C

I think I saw conversations back then about a year ago about the file type and so like actually for I.

D

Know I understand now the.

C

Rationale for doing a slash ipfs in the hash and before, but in order for it to be a properly included URI. If it would be like a file, colon, slash, slash, slash or.

A

C

Colon slash, slash and so I think now, realizing it sort of ipfs really as a file system and so I think, really that it and I think it changed. The challenge would be for the I didn't mention this, but actually within the context, is for the Jason that lead parsers. Actually, they recognize that it's a it's an IP FS hash and then actually to recognize those characters in the string. Then, actually it been to resolve the location of the context, yeah.

D

C

I think again, these are early days and I think we should have. This could reduce yeah.

B

Anyway, it was.

C

B

C

B

This this for everyone to join and for I, want to raise questions, and this this time was a very quick update. That I mean me Richard quickly at home okie-dokie.

B

This is the end, and it was the good coaster and I would like to thank you. Everyone and I'm flying now so have a good holidays. Everyone bombur's.

A

B

A

B