Directly synced with Varun. Looks good. Before landing make sure to remove need for unnecessary partition key check and add a comment for changes starting from line 340 in services/identity/src/database.rs

Rebase

Address review

In D12673#358848, @marcin wrote:

I am afraid that there is a vulnerability in this diff. Basically it is possible that two createOlmSessionWithPeer promises will execute concurrently. This might lead to corrupted state. In order to mitigate that I think it is best expose this method as a context. The context would internally keep a map like [(userID, deviceID)] : Promise mapping from particular deviceID and userID to currently executing promise (if any).

Shouldn't the test plan be expanded?

update

update

For now it is possible to add any user that shows up in search results. On native we handle this in the UI by showing a popup if the user wants to add someone they cannot:

Use new approach - see the discussion in ENG-8710 - we want to unify web and native

address comments

Going to land this without review in order to unblock the build

Rebase

Adding @will and @varun in case either of them is up right now

rebase before landing

rebase before landing

rebase before landing

address comments + rebase before landing

Thanks for explaining

rebase after resequence

rebase after resequence

Was writing inline comments explaining why fid && fid !== NO_FID_METADATA can be just fid

resequence

address comments

In D12601#357386, @ashoat wrote:

I think it's better to have type suppressions than to have to maintain the old set of types indefinitely. This makes sense to me

JK, the ESLint issue was in revision one before the one that landed. Re-landing.

re-open to address eslint thing