As we continue to iterate on and refine the browser, we’ve been leveraging the community to help us squash bugs. While we’ve had a lot of great user reports from the community that have lead to bug fixes, it’s a lot to ask our users to specify everything about their browser, try to reproduce the issue, copy their console output, etc.
We could silently log errors as they happen, but we want to respect user privacy, and prevent potentially sensitive information that may be hidden in the error from being sent to a remote server.
Implement an error screen on the browser that catches uncaught errors, and presents them to the user for approval. Allow the user to read and approve the error before sending it to us, and also add a field for the user to describe how they got the error, and any steps to reproduce. They’ll have the choice to send it to us, or they can decide not to and simply refresh.
The report will be sent to our API where we will only record the essentials (The error, user notes, basic browser information) and discard any identifying information (IP address, user information.)
- Is there any information we think absolutely shouldn’t be sent that I didn’t mention?
- Should we ask for more / different structured information? Contact information to follow up?
- Should we ask / give the option to join our slack to discuss the issue?