search

ipfs-inactive / sig-blockchain-data

[ARCHIVED] Using IPFS for storing data for Blockchain apps.
32 stars 10 forks source link

Welcome #1

Open jbenet opened 9 years ago

jbenet commented 9 years ago

Carrying over from https://github.com/ipfs/faq/issues/14 (figured a repo with multiple threads made more sense)

jbenet commented 9 years ago

@shea256 @muneeb-ali @compleatang @prestonjbyrne @buchmanster @taoeffect

jbenet commented 9 years ago

what's Devon Read's github?

prestonjbyrne commented 9 years ago

@ebuchman

jbenet commented 9 years ago

@guylepage3

jbenet commented 9 years ago

@john-light

prestonjbyrne commented 9 years ago

@androlo @dennismckinnon

jbenet commented 9 years ago

Everyone, it sounds like all our work is converging faster than expected. Let's discuss here-- it's open, public, and has much more space than twitter.

jbenet commented 9 years ago

@muneeb-ali said

Some background on the data-storage of Blockstore:

https://github.com/openname/blockstore/wiki/Data-Storage

muneeb-ali commented 9 years ago

Yes, so I'll be adding more info/diagrams on the data-storage page. @jcnelson has started looking at a whitepaper today that will also discuss this (and we can try to keep the Github page updated with the whitepaper)

shea256 commented 9 years ago

hola

jbenet commented 9 years ago

watch this Repo

DevonJames commented 9 years ago

'ello ;)

whyrusleeping commented 9 years ago

checking in for updates

shea256 commented 9 years ago

Here's the proposal for blockstore to support the packing of many files into a single Bitcoin transaction: https://github.com/openname/blockstore/issues/81

Would love all your thoughts on this. @DevonJames this is especially relevant to your concerns about high fees.

jbenet commented 9 years ago

@muneeb-ali @shae256 it would be useful to setup IPFS as a blockstore mirror sometime soon. You can use it as a key-value store.

shea256 commented 9 years ago

Let's do it. Also watch out for that typo!

muneeb-ali commented 9 years ago

Adding a little historic context to this conversation. The DHT community in the early 2000s got really excited about things like replacing DNS, replacing lookups on the Internet, decentralized storage, and so on. But none of that really materialized.

This is the "second wave" in a way, to take a stab at these problems and we should be aware of what is different now compared to earlier. One thing is obviously the blockchain ...

DevonJames commented 9 years ago

we're all just updating ourselves on IPFS over here, and really quite stoked

DevonJames commented 9 years ago

changes our thinking about a lot of stuff - primarily how much info should go in the chain since IPFS looks to be way faster than the mainline DHT

DevonJames commented 9 years ago

@muneeb-ali ,@shea256 - since your time with onename and now working on blockstore, whats your take so far on the specialization of namecoin vs security (even if it requires somewhat tweaked mechanisms) of bitcoin?

DevonJames commented 9 years ago

oh, and @jbenet can you explain how filecoin and IPFS relate to one another?

DevonJames commented 9 years ago

we're really weighing the question of a specialized alt coin vs changing data structure to use BTC blockchain, but IPFS as a replacement for mainline DHT makes it more compelling (much faster discovery and loading) but then how does filecoin play into the mix?

DevonJames commented 9 years ago

plus, of course, this: https://github.com/openname/blockstore/issues/81 to reduce the cost of submissions

ebuchman commented 9 years ago

couple things: 1) would love to see a standard/initiative emerge to get a single DHT backbone for bootstrapping all p2p protocols. We could eg. use DNS to map a url to some "protocol id", and then find peers for that protocol on the DHT. You know, put an end to the bootstrap problem once and for all. We might need to pester OS/browser vendors to get onboard so we can easily bootstrap onto the DHT itself. A "one bootstrap, one time, to rule them all" approach

2) I'm very interested in building a minimal blockchain in javascript entirely on top of ipfs. That is, put a directory of js files and a genesis block on ipfs. A user pulls the js, its presumably a node app, say, it could even be python we could do it 3 languages as a proof of concept. then all blocks/txs could just be ipfs blobs. the trick will be how to handle the non-persistent network messaging, like for getBlocks or getTxs. @jbenet does IPFS support any form of custom overlay networks so I can write a p2p protocol above it?

I'm still in SF. Maybe we can plan a few days to get together a few of us and try and hack on something like this?

whyrusleeping commented 9 years ago

@ebuchman you can build an overlay network on ipfs very easily. I have some examples here: http://gateway.ipfs.io/ipfs/QmTkzDwWqPbnAh5YiV5VwcTLnGdwSNsNTn2aDxdXBFca7D/example#/ipfs/QmX6gcmX2vy2gs5dWB45w8aUNynEiqGhLayXySGb7RF2TM/readme.md

DevonJames commented 9 years ago

@jbenet @prestonjbyrne thoughts on webtorrent? turning users on websites into seeders could really be a big deal. https://instant.io/

whyrusleeping commented 9 years ago

@DevonJames we plan on having an implementation of ipfs in javascript so we can do exactly that!

jbenet commented 9 years ago

@ebuchman

would love to see a standard/initiative emerge to get a single DHT backbone for bootstrapping all p2p protocols.

entirely agreed. it's not trivial because DHTs can be hard to get right for each different use case. for example, we didn't use MainlineDHT because it wouldn't be nice to torrent clients. we dump lots of records into our DHT.

@jbenet does IPFS support any form of custom overlay networks so I can write a p2p protocol above it?

as @whyrusleeping said, yep. can mount your own protocols on top of ipfs. can use it as a libp2p of sorts. (we actually plan to extract this out in time).

I'm still in SF. Maybe we can plan a few days to get together a few of us and try and hack on something like this?

Happy to. i'll be in SF wed and thu for

jbenet commented 9 years ago

@DevonJames

can you explain how filecoin and IPFS relate to one another?

Filecoin is a sister protocol of IPFS. you can use either without the other. We're focused on getting IPFS solid first.

IPFS as a replacement for mainline DHT makes it more compelling (much faster discovery and loading)

Yep that's why we made it.

muneeb-ali commented 9 years ago

@DevonJames wrote:

@muneeb-ali ,@shea256 - since your time with onename and now working on blockstore, whats your take so far on the specialization of namecoin vs security (even if it requires somewhat tweaked mechanisms) of bitcoin?

I've added more notes on the choice of blockchain here: https://github.com/openname/blockstore/wiki/Blockchain

Let's try to define how exactly are we comparing the different blockchain options (what metrics do we care about) and then it'll be easier to compare Namecoin vs Bitcoin vs something else.

As you can probably guess, out of the metrics I've listed. The Bitcoin blockchain clearly outperforms Namecoin on every single metric.

ebuchman commented 9 years ago

@jbenet

it's not trivial because DHTs can be hard to get right for each different use case.

Sure, and that's fine, but I'm talking about a DHT for a single common use case of finding peers for every other DHT. So eg someone wants to connect to IPFS. They know the ipfs protocol url. Say its "protocol.ipfs.io". DNS query returns a hash. Then they look up that hash on this global dht to find IPFS peers. then they can connect to those peers and join IPFS' dht. and so on. Would be really nice if some international standard was set for bootstrapping seeds in p2p protocols is all.

can mount your own protocols on top of ipfs.

Awesome.

I'll try and swing by tmrw/thurs.

jbenet commented 9 years ago

@ebuchman

Sure, and that's fine, but I'm talking about a DHT for a single common use case of finding peers for every other DHT. So eg someone wants to connect to IPFS. They know the ipfs protocol url. Say its "protocol.ipfs.io". DNS query returns a hash. Then they look up that hash on this global dht to find IPFS peers. then they can connect to those peers and join IPFS' dht. and so on. Would be really nice if some international standard was set for bootstrapping seeds in p2p protocols is all.

very much agreed :+1:

denisnazarov commented 9 years ago

Wanted to add https://github.com/mine-code/canonical-content-registry to the conversation.

CCR goal is to write a lot of metadata to a DHT timestamped with the blockchain, with media retrieval as a necessity down the line.

@moudy @jessewalden