Rebase on a master changes.

In D5166#151293, @jon wrote:

Unfortunately, I don't have integration tests added yet, see https://phab.comm.dev/D5164#151290 .

related: https://linear.app/comm/issue/ENG-1814/distinguish-between-unit-and-integration-tests-for-tunnelbroker

Rebased on parent changes.

Rebased on parent changes. Error message was fixed.

Rebased on changes. Default false value was added for an online status on a new session insert.

In D5173#151887, @jon wrote:

In D5173#151704, @max wrote:

I'm checking what's wrong with the Android and iOS builds (maybe the protobuf changes broke some old client-side API code). Requesting review to unblock it for possible comments.

We currently have set(CMAKE_CXX_STANDARD 14) in the CMakeLists.txt file, however, this commit now includes C++17 extensions

In D5179#151896, @jon wrote:

I feel like having a test case for this would help a lot of asserting that it has the desired behavior

I'm checking what's wrong with the Android and iOS builds (maybe the protobuf changes broke some old client-side API code). Requesting review to unblock it for possible comments.

I'm checking what's wrong with the Android and iOS builds (maybe the protobuf changes broke some old client-side API code). Requesting review to unblock it for possible comments.

I'm checking what's wrong with the Android and iOS builds (maybe the protobuf changes broke some old client-side API code). Requesting review to unblock it for possible comments.

In D5164#151290, @jon wrote:

Passed a unit test in D5166.

Unfortunately, DatabaseManager tests are being filtered out because there's a requirement on external services. I don't have integration tests yet added.

Nice catch @jon ! I should add env_logger to the Tunnelbroker Rust lib as well.

Change the interval to 3s accrording to our keyserver interval.

Abandon due to ENG-1764: Distinguish common and wrong/expired device token errors in Rust notification library.

Abandon due to ENG-1764: Distinguish common and wrong/expired device token errors in Rust notification library.

Pinging @ashoat for a final blocking review because this is a dependency change.

Tunnelbroker already has ConfigManager abstraction from which we are handling options. It uses boost::program_options as well as in your solution. I think we should use the current approach instead of parsing CLI elsewhere. In this case, we have a single point of truth for options and a single place where it resides.

In D5064#149334, @jon wrote:

This isn't necessary on the nix develop side, but this probably fixes the docker build as it has an older version of openssl available.

That makes sense! We don't need an arc in lazy_static here.

Adding topic for APNS configuration, because it's mandatory in production mode.

Rubber stamped! I'm commenting on this line every time to fix the Clang conflict in Nix while using Clang from XCode.

Rebase and merge.

Rebase and merge.

Android build error is unrelated to this diff (it doesn't touch any Android stuff), it fails with the downloading timeout.

Rebase on master and merge.

In D4540#146009, @atul wrote:

In D4540#146008, @max wrote:

In D4540#145965, @jon wrote:

Android build fail isn't related to this diff:

You should be able to click on the gate, then select "restart build" on the right hand side.

Oh, I thought the state in phabricator will not update on this (it's linked to some unique id of the build run). That's cool, thanks!

The tricky thing here is if you restart a job from the Buildkite UI it won't be reflected in Phabricator, but if you restart a build from Phabricator (via Harbormaster) it will be reflected

In D4540#145965, @jon wrote:

Android build fail isn't related to this diff:

You should be able to click on the gate, then select "restart build" on the right hand side.

Rebased on parent changes.

Android build fail isn't related to this diff:

error An unexpected error occurred: "https://registry.yarnpkg.com/setprototypeof/-/setprototypeof-1.2.0.tgz: Request failed \"503 Service Unavailable\"".

Removing of wraping result in Ok.

Removing of wraping result in Ok.

Rebase on the parent changes.

Use Anyhow and Result instead of bool.
Removing unwrap().
Using snake_case in Rust code, export to C++ as a camelCase.
Simplify the code.

Removing @varun and @ashoat because of the vacation.

In D4540#142224, @varun wrote:

Can we create an Error type for this library and avoid panicking? It's fine in the lazy_static block, but I don't like the use of unwrap elsewhere

Edit: Alternatively we can use anyhow. See this example: https://cxx.rs/binding/result.html#returning-result-from-rust-to-c

Using Anyhow and Result instead of bool as a return.
Remove unwrap()ing.
Using snake_case in Rust code, export to C++ as a camelCase.
Simplify the code.