Notes: BigchainDB IPLD Collab

[jbenet]

@TimDaub here are our TODOs from https://pad.riseup.net/p/Ctdc9oGWjy1Q

TODOS:

• [ ] Juan to standardize (on path to IETF/W3C)
• [x] Juan to figure out multibase
• [x] Tim to connect Juan and {Dimi,Rodolphe,Alberto}@bigchaindb.com
• [ ] IPFS team to write py-ipld - https://github.com/ipfs/py-ipfs/issues/29
• [ ] IPFS team check completeness on py-ipfs-api
• [ ] BigchainDB:
  
  
  
  • [ ] use IPLD and multihash
  
  
  • [ ] resolver for IPLD over BigchainDB
  
  
  • [ ] bigchaindb.resolve(ipld_hash)
  
  
  • [ ] CBOR in BigchainDB / RethinkDB
  
  

[TimDaub]

@jbenet https://github.com/bigchaindb/py-ipld/issues/1

I did a port of js-ipld to Python yesterday.
Currently finishing setting up the repo on Github.

[TimDaub]

There is a very experimental branch on BigchainDB using IPLD now: https://github.com/bigchaindb/bigchaindb/pull/381

As far as I can tell, we'll probably make it an option to enable in a configuration.

[jbenet]

@TimDaub Excellent news!

[jbenet]

Hey @TimDaub wanted to check in here. Some updates on our end:

• We're working on multihash rfc style spec.
• We're pushing fwd multibase too. we're implementing in the next 2-3 weeks.
• Looks like py-ipld is good to go? anything else left there? what do you need from us?
• Working on IPLD spec too-- pushing fwd a spec for either W3C or IETF.

We're pushing hard on IPLD this quarter, and putting it into go-ipfs and js-ipfs. I think it would be useful to iterate together more closely these coming weeks. What do you think of a regular weekly 30min call to check up on things?

[TimDaub]

We're working on multihash rfc style spec.
We're pushing fwd multibase too. we're implementing in the next 2-3 weeks.

Awesome!

Looks like py-ipld is good to go? anything else left there? what do you need from us?

Nope, py-ipld is basically done. Huge thanks to @dignifiedquire!
I guess what could be a missing link at one point for us (2-3 months into the future) is a full-fledged Python IPFS implementation. On Friday @gmcmullen and I have been sending out the first COALA IP specification draft. We'll be using that as the main driver for a lot of our efforts on BigchainDB. One element of it is that some media blobs (specifically contracts in forms of .pdfs) will have to reside on a content addressable storage solution (IPFS in essence). Just wanna make sure that this will not be a blocker eventually.

Signed/authenticated IPLD objects might be another potential blocker but the discussion about it gladly already started: https://github.com/ipfs/notes/issues/152#issuecomment-239320116

Working on IPLD spec too-- pushing fwd a spec for either W3C or IETF.

Awesome. I also saw that you've been restructuring a lot of your Github repository. Looks all really nice. Like the work that's being done on IPLD. Seems super promising.

Maybe a quick update on our site:

We've now fully initialized the implementation of the COALA IP.
We're currently working on:

• Implementing a Python COALA IP client that can serialize and deserialize COALA IP models
• Implementing a COALA IP ledger plugin for writing COALA IP models into BigchainDB (in the future there can be one that writes into IPFS ;))
• Implementing IPLD into BigchainDB

I think it would be useful to iterate together more closely these coming weeks. What do you think of a regular weekly 30min call to check up on things?

Makes sense to me. I'll write you an email about scheduling.

/cc @sohkai @sbellem

[TimDaub]

More updates :tada::

• We're getting closer to IPLD integration: https://github.com/bigchaindb/bigchaindb/issues/485#issuecomment-251714485
• We're making really good progress on implementing the Python COALA IP reference implementation: https://github.com/bigchaindb/pycoalaip

[TimDaub]

We also _finally_ released a first draft of the COALA IP specification: https://github.com/COALAIP/specs

We're using IPLD for linking in-between documents.

[jbenet]

Hey @diasdavid and @dignifiedquire -- talking with @TimDaub in person and brought him up to speed with CID and IPLD and new things. They:

• can use CBOR (and use the cbor stuff we have)
• or keep their stringified json (and make ipld-dag-json or something)

Either way, they'll need your help to integrate all this and make sure it works well.

[daviddias]

Sounds good, let's make this happen!

Since your data is already in JSON, you should be able to add it directly using dag-cbor, I believe it is the best bet because it will save you from some future annoyances and considerations like supporting binary data in your JSON blobs.

Can you provide me some data that you currently store in BigchainDB and how it is being linked?