rebase and alnd

In D11209#324215, @tomek wrote:

However, if there is no specialRole field encountered, we will assume we're dealing with a Server/"legacy" type and fallback to sketchy string search.

What is the scenario when this can happen?

memoize useRoleNamesToSpecialRole

Removing Ginsu as reviewer since he's OOO and feedback was addressed.

Will run on the MacBook Air (where it was previously failing) here: https://phab.comm.dev/D11245

setting JAVA_HOME before command worked, using the env approach with ANDROID_SIGNING_PASSWORD didn't.

bring back mkdir w/ -p flag

Forgot this was landed, looks like it brought iOS build times down to ~13-14min from ~50min

Version will be bumped in "flip the switch" diff. native, web, and keyserver all need to be updated at the same time as a release.

update

rebase and land

Looks good, just want to make sure that this won't break any older clients? I see the following in Test Plan:

Looks good

I don't have any experience with web workers, will defer to other reviewers

Mostly looks good, but do we need to handle clearing the data on log out/delete account/etc?

Looks like the Emscripten workflow is failing? Sending back to your queue, feel free to re-request review if I'm missing something

Passing two command flags seems weird, can we do like yarn build-db-wasm && yarn build-db-wasm?

In D11099#320649, @tomek wrote:

What is the downside of making this diff a draft instead of marking its title with a label?

tmp

update and land

update and alnd

No longer relevant since we have separate server and client types. In the future we will flip things when we have minimally encoded permissions in memberships table, at which point we can change the logic here.

Looks good. Also looks like we don't need to run yarn build-db-wasm on these changes

stage all

land

Sorry, added annotation showing original code.

land