Another typo

Fix typos

Remove extraneous newline

Probably need to pair with someone to assert that this is working from a clean state.

I don't think this is right. Can you update your test plan to include registering / logging in / playing around with / refreshing the app? (Also the landing page)

Refactor into script and nix wrapper

Refactor into script + nix wrapper

I'll defer to tomek for final yes. But I don't see anything wrong

defer to someone else for js

What downsides do you see to that approach?

Can we stop writing bash inline in Nix files? This has been discussed numerous times during review, not sure where we landed but it still feels like a code smell

Only partially related to this change, but we should avoid assuming configuration files will be co-located to the binary. Instead of defaulting to a relative directory (e.g. ./secrets/), we should really be defaulting to something like XDG_CONFIG_HOME which is usually only writable to a specific user. For linux systems which have adopted KDE, GNOME, or systemd, these values will be well-defined for users and services. https://specifications.freedesktop.org/basedir-spec/basedir-spec-latest.html. The other benefit is that the file location will also not be dependent from where you run the application. (e.g. cargo run vs cargo build && cd target/debug && ./identity vs cargo install && identity)

An antipattern in C++, that should be avoided, is using using.

I'm in the process of pushing up the new builds for macs to the cache

I'll defer to tomek for approval, I don't see anything "wrong" with this though.

It's still taking a while to install nodejs-16.13.0, has that not been added to the cache yet for aarch64-darwin?

I mean that's just Identity tests and Tunnelbroker unit tests, right?

looks fine to me, but haven't used java in 10+ years.

I'll defer to others for a blocking review, but doesn't look wrong to me.

doesn't look wrong, but not familiar enough with jni to know if it's right. LGTM

Abandoning revision, as we likely want to do an office hours around how this data should be modeled

karol left

looks like I was added because the initial diff made some changes to native/cpp, going to resign, as I'm not qualified to review this

We are pinging to detect when the client goes offline so this isn't an exact error in our case. It makes sense to log it only for debug purposes using the INFO.

Feel like there's probably a more elegant way to do the Option<Hashmap<,>> logic, but nothing comes to my mind.

LGTM, as long as https://phab.comm.dev/D5207 lands at the same time.

LGTM, but don't consider myself familiar with C++ best practices, so I'll defer to someone else for marking this as accepted.

Assuming that the new token would occupy the responseMessage field in the response, this looks good to me.

We can always revisit if we want more detail than just a bool to denote success/failure. But that decision can be deferred.

looks like the awkwardness is for Rust + C++ compatibility, LGTM

LGTM, I like it

I don't know if this will protect against sockets which close (which I think is the underlying issue for most of these failures), but may as well try

The prompt issue was fixed with https://phab.comm.dev/D5194. For CI we need to "give permission" ahead of time to receive the comm binary cache. The default behavior is to prompt a user, but for CI, there is no user.

Best to leave me off of these reviews

Apply --accept-flake-config fixes, run if nix files change

abandoning in favor of https://phab.comm.dev/D5181

not familiar enough with the sqlite_orm to have meaningful feedback.

In the future, I would give the directions from the perspective of "I just did arc patch", for example:

cd services/commtest
cargo test --no-run

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