IndexedDB proxy

If your game relies on IndexedDB in order to persist data, this bridge will allow it to work in Kate by converting IndexedDB operations to work on the Object Storage API. You need to choose whether you’ll use the versioned or unversioned partition when instantiating this bridge.

Using in your build configuration

To add this bridge to your cartridge you specify the following in your kate.json configuration:

{
  "bridges": [
    {
      "type": "indexeddb-proxy",
      "versioned": true
    }
  ]
}

The versioned option defines whether the bridge will use the versioned partition (where data is only stored for the current cartridge’s version), or if it’ll use the unversioned partition, which is shared between all versions.

Note that IndexedDB is designed to support schema upgrades, but not schema downgrades. Kate players expect both to work, so the disabling versioning is only a good option if you can guarantee that your database version will never change. Otherwise players will need to delete all their save data in order to be able to downgrade your cartridge.

Dynamic emulation

This bridge works by emulating the IndexedDB API dynamically. It replaces the global indexedDB object with the bridge’s own implementation of the API on top of Kate’s Object Storage API.

However, there are problems you may hit with it. Because the IndexedDB API is quite substantial and complex, this bridge only implements a subset of the API that is more likely to be used by a video game. It has enough implemented to work with Emscripten’s IDBFS implementation (which powers e.g.: Ren’Py’s save data feature on the web export).

It also lacks any transactional guarantees and any optimised implementation of indexing and searching (the implementation always performs a linear search over the entire table). This makes the bridge a poor choice to anything that needs to store and search tens of thousands of items, or which needs concurrency and transactional guarantees.

Note

If you have a use case that is not covered by the current proxy implementation of IndexedDB, please open an issue on GitHub to discuss the specific scenarios you’re looking at. A full implementation of the IndexedDB is very unlikely, but covering more common cases is desirable.