Skip to content

Within QE what is the equivalent conception of a folder or dataset?

  • A recursive key value hierarchy where each key value pair has RBAC
  • Data sources are traditionally DNS based, I remember installing packages as com.google.PACKAGE_NAME, CGFS also require labels. Just like with Persona that are public keys datasets are can also be seen as data signed via a public key

Brainstorming

  • 2024-01-29
    • We just need some CID that can fit into a a special datatype memelet.
    • No what we want is a IPNS situation
    • Ah people need their own RBAC name spaces
    • We can think of people as their own unique Domains, reminds me of LDAP
    • But people like to rename their namespaces
    • We can use the root CID stating that "MY LIFE HAS VALUE" as a root, but this CID does not even need to exist
    • Okay give me a use case
      • I have my YouTube Playlists as a dataset
      • I also have my youtube watch history
      • I also have videos I uploaded to youtube
    • Alright what do you want to do with this data?
      • I want is accessible, like I want to select my most recent videos from youtube
      • Okay what is that good for
      • Dude not now
      • I need DNS type recursive name spaces
      • The name spaces are fluid and attach to different root ID's
    • Ah are we really dealing with the name space problem
    • BRO this is solved via Google Plus Circles, and circles work with DNS
    • So basically a recursive key value hierarchy where each key value pair has RBAC
    • Wait a file system is basically a key value hierarchy
    • Ya that's exactly what it is
    • Are there any other examples like this?
    • Well we need intersecting key value pair systems
    • That's where the CID's come in, the same file can be referenced on any system
    • Does IPNS support a key value pair system?
      • Oh shit this basically solves my exact problem
      • IPNS just allows private keys to act as their own self sovereign domain name as long as they run their own IPFS node just like a DNS server
      • Also uses DAG-CBOR to support a key value store
      • There is a limit of 10 Kib for the record list, reminds me of npm max of 800 dependencies
    • Ummm so this IPNS spec actually looks perfect, maybe 10 kb limit is a problem but anything like that should be something else
    • Or we can just ignore the IPNS limit by using some library that implements the data because this is not supposed to be broadcast on IPNS but used locally, we just want to reduce the number of standards we use
    • Well we need a key value store and this is a really nice signed key value store
    • Ya I guess we can use it, now can we find a library that uses it
    • Is there a IPNS JS library?
      • Yes but not for extendable data fields
      • Hmm I like the idea of every directory just being a name space managed by their own private key that signs data
      • That's going to be a lot of private keys
      • It allows for sharing that involves real trust
      • Ya I like this let's see if we can do it
      • I could only find Snapshot IPNS names in my ENS Indexing data
    • How to create your own IPNS name?