Fix typo

I don't think we should land this if we're not going to put it into CI, and I don't think we can land this with a CI job since people don't really have a way to iterate on it / test it.

Remove docker image

I'll scope this down to remove the docker image, it's a nice-to-have. It will be easy to add it back.

More consistent formatting and prose

Fix grammar

Nevermind, release mode removes developer tools section from profile.

How are you making sure that this doesn't mess up the config in our production environment? Can you update your Test Plan to include steps to address that?

Do we have to assume that devs are using the nix supplied mariadb?

Eventually, I would like to get the docker images to also use the same protoc as the rest of the project, ubuntu has been on 3.12 for the past ~4 years.

Do you foresee any need for developers to test building the Docker image locally?

I agree with @bartek on this one. The existing docker solutions work together for also the dev environment. We should update how those work first as that's that only environment which we have integration tests working currently. Once https://linear.app/comm/issue/ENG-1697, https://linear.app/comm/issue/ENG-1696 land, maybe https://linear.app/comm/issue/ENG-1792; then we can revisit the options, as we should have a "purely" native way to run the service test suite.

Add restart of command

Rebase on master

Can you amend the Test Plan to include testing this on macOS, both Apple-based and Intel-based?

Use more consistent wording for user

looks pretty standard to me

Seems like interesting way to go from &mut to an owned version of itself; as long we no longer need to access the chunk after we "reset" it.

Should also remove the corresponding comm::network::tools::isSandbox() method if it's going to exist within the config file.

Maybe we can unify this in a way that comm-dev services stop stopping or/and killing services and we just call comm-dev services stop from the inside of comm-dev services start and move all of the "stopping" logic from the beginning of the comm-dev services start to comm-dev services stop?

This looks right to me

Remove commSrc as it rebuilds everything. Fix typo

Should we abandon this now that shared/protos/CMakeLists.txt isn't used from the Android build following D5706?

Add back darwin fixes

Don't expose docker image on non-linux platforms

It looks like the image is built for a different architecture than mine. Am I missing something?

@jon it would save the whole team SO MUCH time if you started testing ALL OF your diffs on the same environment that literally everybody else uses: macOS

The better solution would likely be to get tunnelbroker to use protobuf 3.15 rather than 3.12

Can we update the Test Plan to include what operating systems and architectures the diff was tested on

not familiar enough with c++ best practices to provide valuable feedback

I feel like this would be a good candidate to include a unit test. Also demonstrates the ergonomics of "using" the hash method we just introduced.

I think the object oriented way is a bit cleaner, especially if everything needs to go through a context object anyway.

Fix darwin build

I don't see any other smells, but going to defer to @varun for final say so

This was meant to show a rough level of complexity that would be involved with packaging rust comm services with Nix, wasn't sure if we wanted to travel down this path.

Add default user and pass

Merge conflicts

I just yarn arcpatch'd this and found that:

Removed every non-Nix protoc and things still don't work:

Do we need to update something on CI side?

I believe the intent here was to remove "you can" because it could give the impression that a step is optional.

Fix rebase

Additional fixups

Make older shellcheck happy

Refactor most of the logic to be a shell script

I think the intent was to avoid an imperative or ordering tone. Probably want something like Please run the following: instead of Run this command:

It's still not working and it's pulling the following Protobuf:

Reduce scope to just grpc code

looks to be pretty standard way to use tracing

Still build fine, so not likely to be used.

Overall looks good