IPFS Virtual Meetups, 30 Apr 2021

Previous Meeting Next Meeting

⏯

youtube image

►

From YouTube: NFT x IPFS Best Practices with Yusef Napora

Description

Yusef Napora, Protocol Labs software engineer, covered NFT x IPFS best practices, and introduced us to Minty at this month’s IPFS meetup.

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

Join your local IPFS meetup to attend our next event: https://www.meetup.com/pro/ipfs/

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

A

Ipfs and nft best practices.

A

uh What are those right, and we've spent some time thinking about this and we've kind of come up with a few recommendations that we think will ensure that your nfts end up in a good state, because there's a lot of like subtle missteps you can make along the way and because nfts are pretty permanent. You kind of want to get it right on your first go. If you can uh so, but first before we get into like you know, it's ips, I yeah. That's specifics!

A

I'm gonna talk about like what actually is an nft right because, like it's not this image over here, although there is an nft of this image, but you know, of course anybody can grab these pixels and stick them in a presentation or whatever it's that so the the media itself is not really the core of the nft in almost all cases, um but what what it is is like a a record of ownership, that's public on a blockchain, so these there's a few like when you boil it all down like what actually is the nft.

A

It has a unique identity.

A

So, there's only one specific one of each token: you can't trade them about they're non-fungible, so you can't like exchange one for another, an arbitrary way and then you've got they can be transferred to a new owner and owned by a specific account on the blockchain, and they can generally only store a tiny amount of data directly on the blockchain I mean this is not strictly speaking, like a you know, specific to nfts themselves, like you could imagine a blockchain where you could store infinite data, but today's nfts are very limited in terms of this on-chain storage space.

A

So they have this. You know escape patch, basically, which is like you can link to data off-chain using a uri or universal or uniform sorry resource identifier, and this is the primary method of actually getting the nft to mean something as it were so right, an nft generally when people think of it. They're like oh, it's this cool artwork or it's it's music or it's whatever, but on the blockchain itself. All you really have is the uri that points you to the thing. So here we are at uh our public service announcement.

A

Friends, don't let friends mint nfts with http links, because uh here we yeah they are fragile, although they are super familiar and easy and they're widely supported. So it's easy to understand why you might want to use them, um but they can change at any moment.

A

Like I mean I can write, a web server now returns the current time and every time you request that url, it's going to return something different, um and the same is true for any artwork that you might want to put your or any asset that you might want to make an nft out of the the url points to a location and that, what's at the other end of that location can be anything and so there's no guarantee that it's going to be the same thing that the nft creator put there.

A

So we have ipfs, as most people here will probably know like ipfs uses content addressing and it basically now you have a link that actually can't change right. If the uh the ipfs cid your content id will only ever point to the specific piece of data that created it. So we think that this is the best way to refer to data from a blockchain, basically, because it will always point to the thing that you thought it was pointing to when you created the record um and on the retrieval side, you get built-in verification.

A

So when your ipfs client fetches the data, it's going to compare the hash to the cid, if they're not a match, it just won't. Give you any data. So there's no like real danger that you're going to be like. Have the rug pulled out from under you as it were, your your all. Your goodies are changed for something else, um and it's also location, indepen independent. I should say so.

A

um You don't have to plan at a specific server that could go down or change ownership or whatever you just say, here's the content and go find it on ipfs, please, and that it makes it so you're not really tightly coupled to any particular host, um which we think is a really powerful advantage. um And then you can also sort of be served over http. In fact, a lot of people use ipfs without really realizing it. Just by going through ipfs gateways, you follow, you click a link to ipfs.io or dweb.link or cloudfare's gateway.

A

There's a bunch of gateways that you can use and they'll just go and do the ipfs bits for you and your web browser doesn't know the difference right. So that's great! We don't want to store those links on the blockchain, because you know there's a bunch of gateways and none of them is the canonical gateway and also yeah. They could disappear. Maybe who knows like it would be sad if your nfts were.

A

You know inaccessible now, even though they're still on ipfs, because you pointed somebody at a gateway that no longer exists. So what we think the solution is is just use this ipfs uri scheme the nft metadata standards like erc 721. They want you to return a uri and hey. We've got one: it's it's really simple: to construct, one you take the ipfs cid and you stick ipfs colon, slash, slash in front of it and you've got an ipfs uri.

A

You can also optionally have a path component. If you store your data in ipfs inside a directory. Now the cid points to the directory object, but then you can also add the file name has a nice human readable path, which I think is nice, especially inside json metadata that sort of describes the nft where that json metadata can be stored in ipfs. So there's no real size limit. You can pack it full of as many bytes as you want, and human readable uris are nice um and on chain.

A

I still think it's kind of nice to have the human readable file name, but if you're really like trying to save every byte, you might only do the cid without the file name in there that just to point to your metadata, because the way this generally works, I didn't actually make a slide for this part, but there's sort of there's two links in place right. The nft points to a metadata object, which is generally some json and then that metadata object, probably points to.

A

I mean it's: it's unlikely that when you're buying an nft, you you're just buying an nft of json and you're happy with that. Probably what you want is the artwork or whatever. Maybe it's a house or something, but that's going to be described in and referenced in, that json object and so inside the json object. We also think ipfsuri is the way to go, because it's very clear and unambiguous. You get this extra bit of context that this isn't just some random hash from any old system anywhere.

A

It's an ipfs hash. You can use ipfs to retrieve it and here's how and then also um it's pretty easy to rewrite one of these into a gateway url, which is so, if I click on this thing here, it's actually going to be serving it from dweb.link and there you go um so that can happen in your presentation layer.

A

If you're building a platform for nfts, you can write a little tiny bit of javascript, um take an ipfs, uri and point it at your favorite gateway uh or you can do both, because if you just have an ipfs uri people that have grave installed or the ipfs companion browser extension, they can click on these uris directly and it'll all just work either through their local ipfs, node or through their default gateway.

A

If they have one configured but then, if you want to breach the broadest audience like right, when you're presenting the link to the user for them to click on, maybe that's when you rewrite it into an http gateway url, um and then they can also have the yeah okay. This is so. This is me talking about what should you use when probably should skip to this slide a second ago on chain just store the ipfs uri and then return it from your metadata uri function?

A

That is in your smart contract in metadata like in that json bundle again, ipfsuri is the way to go. If you want, you can totally have like another field in there that has mirrors that serve over http or something like that, for you know just for redundancy's sake, but we feel that the ipfs content id is specific to the data is the best like canonical, link and yeah.

A

The uri is a nice way to package that up and then on the web gateway url and if you can't like, if it makes sense to display it to the user, um have an ipfs uri in there that they can use brave or other ips native tools with um and if, as you're, you know a lot of my the recommendations, I'm making to people they're building platforms, but this also kind of applies to people minting nfts, just to kind of like have an understanding in your head of like what do.

A

I want my nft to look like under the covers: it's nice if it has ipfs uris in it right and that's something that you can kind of check out with the nft platforms that you might be working with like do they actually do this in a sensible way or what's going on so um and then there's another thing to mention here about persistence, right um ipfs is you know it's a peer-to-peer content distribution protocol and addressing system and you're as data is requested, it kind of like spreads out through the network.

A

You know when I fetch some data from somebody else's ipfs note. My node is going to keep a copy next time. Somebody wants it. They'll, maybe get it for me instead, but my copy is going to age out right. It's a temporary cache, basically, so that you can't really depend on like ipfs.

A

You can't just put your data onto ipfs at large and just forget about it. Somebody has to be there to provide the data to the network on ongoing basis um and that's you know an ipfs node, that's connected to the internet, running 24 7., and that could be your node. If you want, uh we have ipfs cluster, which is a awesome project for coordinating several ipfs nodes to all, pin and maintain the same content over time, and that in fact is used by pinning services under the hood.

A

Well, if you, so, if you don't want to run your own ipfs cluster, you can run a pinning service um stuff like that. There's other alternatives that um we'll get into later that pl is developing, but the the main the key takeaway is like somebody has to keep the data alive as it were, and host provide it to the network. um So that's you know. Whatever solution you have for that, it's just something to keep in mind because we want, you know nfts to always be resolvable right through via ipfs like.

A

If that's going to be your primary link, then, as a data provider, you just want to make sure that there's always a way to get data out of ipfs and host your own user bidding service.

A

Those are reasonable options and I'll talk about another upcoming option in a minute, um and then here I just want to point you guys at some resources, mostly if you want to really like get into more detail about like why you know these recommendations are the way they are and stuff like that. uh We did a write-up on the ipfs doc site about this a few weeks ago, and you can go check it out at this url down here, there's also at the same site.

A

If you go to mint an nft with ipfs, you can see a little command line, nft mentor app. I wrote, and it kind of you know it's sort of a toy example, but it'll show you the way that, like we think, nft metadata should look and like how the links actually work in practice, and then I also wanted to just call out some really good content on pinata's blog, because they've been producing awesome, nft resources for a while now, and I think it's worth a quick around for anybody that has not seen it yet.

A

So I figured I'd toss that in here as well. um Oh, so this is a sneak preview of alan's talk uh towards the end of this meetup, but pl is launching a new service for nft data storage called nft.storage, and it's awesome and you guys check it out. So, basically, like all that stuff, I said about persistence, well that this is us doing it for you, because we want nfts to be awesome and like we want to preserve the sort of you know: cultural artifacts that are being created and traded and everything else.

A

So I don't want to spoil it too much. I'm sure alan will get under the covers and tell us all about how it works and stuff. But it's it's a very like sort of set it and forget it approach, especially if you're just you, if you have other things to think about you're building an ft platform. Maybe you want to use ipfs, but you don't want to like learn everything about ipfs right now, maybe check out nft.storage, um and that's that's really it.

A

I think that um if you guys, if anybody here has questions for me, uh hit me up, I'm probably gonna give you a faster response. If you email me, uh twitter is something I very rarely check, but I do have a thing on there um and yeah. I think you know as we're learning more about nfts. I'm sure we will have more things to say about this, but I think the key takeaway is like an nft.

A

You can't change it once you're done. If you have a mutable link to like a link that can change, then you've kind of lost that wonderful property of a blockchain, which is that it's in it's immutable and lasts forever, but mutable links kind of undermine that. So an ipfs cid is.

A