twinql

A graph query language for the semantic web. Not SPARQL.

Use Cases

twinql was designed with the goal of scratching a particular itch: fetching linked data over LDP without having to imperatively follow every link and handle every error in an ad-hoc manner in Solid applications.

The main idea behind twinql is that queries select subgraphs by starting at a particular node and traversing outwards. The query and the response have a similar recursive tree structure so that the shape of the response can be inferred from the shape of the request.

It is currently a hobby project and quite limited in scope. It cannot do many of the things that SPARQL can. However, it attempts to be more ergonomic than SPARQL for common use cases.

Examples

Here's how you would query a WebID for profile data and data of that person's friends:

@prefix foaf http://xmlns.com/foaf/0.1/

https://dan-f.databox.me/profile/card#me {
  foaf:name
  foaf:img
  [ foaf:knows ] {
    foaf:name
    foaf:img
  }
}

Response:

{
  "@context": {
    "foaf": "http://xmlns.com/foaf/0.1/"
  },
  "@id": "https://dan-f.databox.me/profile/card#me",
  "foaf:name": { "@value": "Daniel Friedman" },
  "foaf:img": { "@value": "https://dan-f.databox.me/profile/me.jpg"},
  "foaf:knows": [
    {
      "@id": "https://deiu.me/profile#me",
      "foaf:name": {"@value": "Andrei Vlad Sambra" },
      "foaf:img": {"@value": "https://deiu.me/avatar.jpg" }
    },
    {
      /* ... */
    }
  ]
}

Goals

• Be declarative
• Work well with existing standards and tools
• Make app-building easier
• Support multiple persistence layers (in-memory, link following, SPARQL, etc)
• Be implemented eventually on the server to improve performance when querying within a single domain and to reduce data being sent over the network

Known Challenges

• Link-following is slow
• Not all linked data actually links as well as it should
• LDP doesn't care how much data you want from a particular graph; it gives you the whole thing
• Just about no operation on the semantic web is atomic

Roadmap

• ordering and pagination
• response streaming
• Mutation API

Other cool things

• Create higher level tooling for offline-first querying and realtime updates
• Create bindings for common UI libraries
  • e.g. connected React component

Development

This reference implementation of twinql happens to be built in JS for quick prototyping, but a safer language is recommended when implementing for a production use case.

Contributing

If you want to contribute to this reference implementation, first reach out by creating a Github Issue to make sure we're on the same page 😄

Assuming you want to mess with the code, just do the following:

0. Make sure you have node >=7.x and npm installed.

1. Clone the repo

$ git clone https://github.com/dan-f/twinql.git # (or your fork)

2. Install the dependencies

$ cd twinql && npm install

3. Run the demo site

$ npm start

4. Build the lib

# You can run webpack in watch mode to rebuild the UMD bundle on file changes.
# This is useful when prototyping with the demo site.
$ npm run build:dev

# To test the minified UMD build:
$ npm run build:umd

# To transpile the library to CommonJS ES5
$ npm run build:lib