Aaron Boodman

Results 63 issues of Aaron Boodman

Replicache is client-side tech that can work with any backend stack, but we also plan to create our own server that is optimally designed to work well with Replicache. This...

Undo/redo is a very common feature request in collaborative applications and doing it correctly is somewhat subtle (e.g., see posts by [Figma](https://www.figma.com/blog/how-figmas-multiplayer-technology-works/) and [Liveblocks](https://liveblocks.io/blog/how-to-build-undo-redo-in-a-multiplayer-environment)). Many Replicache users have created undo/redo...

Future

It is currently possible to use Replicache cookies and other features to incrementally sync a large client view and/or sync only a subset of available data ("window sync"). We have...

Future
Design Needed

Right now it is theoretically possible to use Replicache with svelte but not easy as there is no sample. It's also possible feature work is necessary to make it work...

Future

Right now it is possible to use Replicache with Solid, but not easy since there are no samples and not efficient since Replicache's `subscribe()` method returns the entire dataset each...

Future

Currently it is possible to use Replicache with React Native by implementing `experimentalKVStore` and hooking it to SQLite. However it's quite manual. We should have first-class support for RN, including:...

Future

People currently implement collaborative text editing by using yjs or fractional-indexing on top of Replicache. We should have a recommendation for how to support collaborative text on Replicache and perhaps...

Future

Replicache currently supports working offline but with some caveats, described here: https://doc.replicache.dev/offline. We plan to remove those caveats so that Replicache can be started offline, work with multiple tabs offline,...

See original request here: https://github.com/rocicorp/replicache-todo/issues/2. This could make sense in Replidraw, for example, some upper limit on number of objects on page or something?

Future

Currently Replicache returns deep clones in all read interfaces for safety. This was introduced in rocicorp/replicache#479 and the rationale is there. However, this must be *really* expensive, both in cpu...

user-reported