Implement analytics on git (git commit)

[loujaybee]

Context: Tracking a git commit is a strong signal that a user is doing something "meaningful" in a workspace, and is useful as the "end" point for our onboarding funnel. A git commit signals that the user has understood things like: Gitpod is a container image, gitpod is mutable and not read-only (e.g. it's not just a https://vscode.dev/ type application), etc.

---

We currently don't know how/when users are discovering terminal access within the browser-based VS Code. This issue is to implement the potential tracking of terminal events within hosted VS Code.

Terminal events that are of particular interest are:

• git commands, specifically git commit as this tells us when a user has acknowledged that Gitpod is not just a read-only application, and that full git workflows are possible.
• install commands such as npm install or apt-get as we want to see if users are understanding that they can codify their installations into the .gitpod.yml or not, and whether they're utilising configurations such as prebuilds.

A technical spike might be required to understand how this could feasibly be implemented. Will also be worth getting the @gitpod-io/engineering-workspace team involved and @csweichel to hep us with this.

DISCLAIMER: We need to be careful with this to respect users privacy. Let's investigate what changes we'll need to make to our privacy policy, or possibly implement an explicit opt-in for users to send us developer data.

[jeanp413]

I think we can use this proposed API https://github.com/microsoft/vscode/blob/main/src/vscode-dts/vscode.proposed.terminalDataWriteEvent.d.ts

[loujaybee]

Spoke with @akosyakov and have reduced the scope on this to be tracking only "git" commands, rather than all terminals.

[loujaybee]

Is dependent upon: #7137

[loujaybee]

/schedule

[roboquat]

@loujaybee: Issue scheduled in the workspace team (WIP: 0)

In response to this:

/schedule

Instructions for interacting with me using PR comments are available here. If you have questions or suggestions related to my behavior, please file an issue against the kubernetes/test-infra repository.

[jeanp413]

From the vscode telemetry:

• event name git.command for all commands in the git extension, it only logs the git command id, git commands executed from the terminal are not detected.

[akosyakov]

We decided to go with tracking git push on credential helper level. Instrumenting git extensions will require maintaining another web socket connection to the server so we rather postpone it.

[bigint]

So is this gonna be an anonymous data collection or do you collect user's data (example: ws id)?

[loujaybee]

Hey @bigint ! Thanks for asking such an important question 😄

Let's move that conversation to this issue: #7317 (comment) as we may need to make some changes to our privacy policy, etc, (e.g. allowing users to opt-in/opt-out of sending telemetry data). From our side, we need some data to better understand how folks are using Gitpod, so that we can do a better job with the product, but we 100% respect our users privacy, and we will be looking into what (if any) legal changes we'll need to make 🙏 .