You fixed it for web but not native (I actually didn't catch web – glad you did)

Separately, wonder if we just want to have a hook like useThreadChatMentionCandidates instead of memoizing the results of getUserMentionsCandidates in both places

In D10387#300593, @ashoat wrote:

You fixed it for web but not native (I actually didn't catch web – glad you did)

Oh sorry about that, I don't know what I was thinking to not do it for both even though it was called out here.

Separately, wonder if we just want to have a hook like useThreadChatMentionCandidates instead of memoizing the results of getUserMentionsCandidates in both places

Yeah that works better - though I was taking another look at getUserMentionsCandidates and I'm not sure if there's a use for memoization here. I don't think this is computationally expensive, but feel free to let me know if you disagree!

function getUserMentionsCandidates(
  threadInfo: ThreadInfo,
  parentThreadInfo: ?ThreadInfo,
): $ReadOnlyArray<RelativeMemberInfo> {
  if (threadInfo.type !== threadTypes.SIDEBAR) {
    return threadInfo.members;
  }
  if (parentThreadInfo) {
    return parentThreadInfo.members;
  }
  // This scenario should not occur unless the user logs out while looking at a
  // sidebar. In that scenario, the Redux store may be cleared before ReactNav
  // finishes transitioning away from the previous screen
  return [];
}

Memoization isn't just about avoiding the performance cost of rerunning the code. What's usually more important is avoiding regenerating non-primitives (like arrays and objects). By avoiding regenerating these, we reduce a bunch of expensive React renders.

In D10387#301064, @ashoat wrote:

Memoization isn't just about avoiding the performance cost of rerunning the code. What's usually more important is avoiding regenerating non-primitives (like arrays and objects). By avoiding regenerating these, we reduce a bunch of expensive React renders.

Got it, thanks for the clarification - makes perfect sense here