Tools for processing and accessing different information types
Executables can be downloaded from https://github.com/commonknowledge/dana-core/releases
Nothing unorthodox here. Clone, run yarn to install dependencies and yarn start to run everything.
If you use vscode, you can optionally clone and run this repository in a development container. The GUI is accessible over vnc on port 5901 (password: vscode).
This project comes with a launch.json configuration for debugging the electron process. If using vscode, you can start this by running the default configuration (usually this can be done by hitting F5.
We have three high-level areas.
src/app– Backend code, running in a nodejs or 'electron main' environment lives here.src/frontend– Frontend code, running in a browser or 'electron renderer' environment lives here.src/common– Environment-agnostic code lives here.
In general, try to keep both frontend and backend code separate from anything electron specifc – if you must use electron APIs, wrap them in an API that could be used (or disabled) in non-electron environments.
The backend is subdivided into domain areas (asset, media, etc). Try to keep the interfaces between these relatively minimal and document them well. In particular, avoid making state changes in one domain that implicitly affect another (for example via database cascades).
The general patttern for communication between domains and between the app and its frontend is to have an interface
file in src/common/*.interfaces.ts.
These should generally have a schema declaration associated with them (we use zod for schema validation) so that client-server environments can do data validation easily. Zod provides
We abstract communication between the frontend and backend into the FrontendIpc interface, which can support
both electron's asynchronous message-passing paradigm and http.
This supports dispatching 'pubsub' events from the backend to the frontend and RPC-style calls. The RpcInterface and
EventInterface types are used to identify specifc RPC calls and events in a typesafe way.
These are documented in src/common/ipc.interfaces.ts.
See also examples in the other src/common/*.interface.ts files.
Archives store their metadata in sqlite documents. We use MikroORM as an ORM to simplify this, which bundles Umzug for managing schema migrations. It's fairly straightforward, but please pay attention to its concept of a 'unit of work' if that is new to you.
Migrations can be auto-generated from entities (assuming the entity files follow the file naming convention of
*entity.ts). To do this, run yarn make-migrations.
The state of the current database schema is tracked in a JSON file in the migrations directory. If you need to revert to a previous version of the schema for generating new migrations, you can take advantage of the fact that this is version controlled to do that.
Please do make sure that you squash any new migrations in your branch before opening a PR in order to keep them managable.
To cut a new release:
- When the version is ready to test, increment the version number in package.json
- Draft a new release. Set the “Tag version” to the value of version in package.json, and prefix it with v. “Release title” should be the same as the tag version.
- For example, if package.json version is 1.0, your draft’s “Tag version” and “Release title” would be v1.0.
- Push some commits. Every CI build will update the artifacts attached to this draft.
- Once you are done, publish the release. GitHub will tag the latest commit for you.
Clearing the cache should fix this
yarn jest --clearCache