feat: memory replica in idb

[Gozala]

Illustration of the design

To try this branch you need to run local toolshed

cd toolshed
deno task dev

And run local jumble pointing at it e.g.

cd jumble
TOOLSHED_API_URL=http://0.0.0.0:8000/ AI_URL=https://toolshed.saga-castor.ts.net VITE_STORAGE_TYPE=cached deno task dev

[ubik2]

If we delete here, we'll pass undefined to the subscriber on line 194, which violates the type signature for that method. It's possible there's some guarantee that we won't have a subscriber in that case, but I'm not sure that's true.

[Gozala]

This is a good catch, thanks for spotting it. I guess I'm not sure what the right solution here, because we need to route this back to sink callback

labs/runner/src/storage/base.ts

Lines 55 to 58 in e07b5c6

	sink<T = any>(
	entityId: EntityId,
	callback: (value: StorageValue<T>) => void,
	): Cancel;

@seefeldb what should I pass into a callback when fact is removed / retracted ? I think it does not happen in practice but I imagine we need to support this. I think we either pass no value or we do not call subscription at all.

[Gozala]

Did this in the meantime to fix spotted issue c899082

[seefeldb]

Let's call it with {} as StorageValue for now. That removes the source and sets the value to undefined. Maybe add a TODO, since this isn't really the same as actual deleting. Thanks!

[Gozala]

Addressed in f648213

[sentry]

Suspect Issues

This pull request was deployed and Sentry observed the following issues:

• ‼️ InvalidStateError: Failed to execute 'transaction' on 'IDBDatabase': The database connection is closing. /:replicaName/:charmId View Issue

_{Did you find this useful? React with a 👍 or 👎}