Alternatively we could use //$FlowExpectedError, but doesn't really matter

As an alternative, in a few places where I had to deal with a deprecated field I typed them as an optional empty field (e.g. email?: empty):

• we don't need to introduce a new type
• flow still guards us against using the deprecated field on the server

Add the python node-gyp fix to all js CI workflows.

Could we amend the test plan with testing on safari and testing of incorrectly decrypted notifications?

We might also want to remove invite secret from urlinfo

Readonly

Additionaly link the init in progress status to the init promise with types, so we won't get this situation in the future (an async operation between setting the status and setting the init promise).

LGTM but I haven't worked with identity service

Will this break logging out on older clients?

Don't we need a migration for web too?

Please check the console.warn/error before landing but looks good.

Nice to see the magic string constants removed!

[...] I guess its due to pt 4 - we cannot check if websocket is empty without hacky ways like timeouts

Update types and client_version_unsupported logic

Do we need to handle socket reconnection?

Should we implement some kind of timeout for the promises?

LGTM although I haven't been really working with react native code lately. Do the diffs for filling in the missing // TODO exist?

Rebase

Rebase migration

Rebase before landing

Setting as "plan changes" because it needs to be defered for a while to let web clients migrate (start using squadcal endpoints instead of web.comm.app).

Replaced by D9598

Nit: I would prefer to see the validators just below the corresponding type definition. IMO it makes it harder for people to forget to update validators when updating the types.

Fix formatting

Rebase

Use genesis.id

Fixes

Thanks for noticing that, not sure how I missed it

This gist contains both a SELECT query based on the UPDATE query and an UPDATE query. Both of these queries show equivalent results under EXPLAIN so I think they should have the same performance. They are both set for my user (3033752 which I took from the table in ENG-5287) and the genesis thread (1).

(indentation fixes, I forgot to commit them)

Indentation and naming fixes

Updated the docs issues, amended the test plan for docker keyserver deploy.