Matt Krick

hey @alexreardon happy to attempt a PR for this one. i noticed there was already an attempt that went stale (#1129). Before i get started any guidance on general architecture...

@philipaarseth FWIW, it's really, truly easier to build your own package to support this. this package has some beautiful patterns in place to learn from (namely snubbing HTML DnD for...

my big concern is that _how_ it works should be irrelevant. the readme should show how to use it, but the internals should be a black box where folks don't...

good thinkin! this is just a dumb client cache. Handling offline stuff should be handled in your data transport layer. if you mix the 2 you get all kinds of...

you can remove the error by doing something like `optimistic(reducer, {maxHistory: Infinity})`, but that doesn't fix the root cause. making over 100 actions while offline seems like a lot. are...

Agreed! No one does it well. I had to PR relay to make it work. FWIW, Here's the pattern I use to manage local/optimistic/domain state: https://github.com/facebook/relay/issues/2481#issuecomment-403649559

@WillSquire i used to do that (local state in redux, domain state in relay) but it's a huge pain. you end up with no single source of truth. sometimes a...

Why not give each action its own `id`? Then just revert the action instead of the entire transaction. Assigning an `id` on a per-transaction basis makes for an all-or-nothing approach,...

With that pattern, won't you have an entity integrity problem? As i understand it, you have a single `loading` boolean somewhere in your state that is turned `true` when something...

yeah, it looks minimally invasive & i don't see a huge performance hit to include the changes. I'm not sure I understand it 100%, but if you write up a...