Cancel Button vs. Timeout Setting: Which preserves repository integrity after aborting a Git operation in Sourcetree?

Goal

The objective is to determine which abort mechanism in Sourcetree—pressing the UI Cancel button or relying on a configured timeout—results in a cleaner repository state after a long‑running Git operation such as fetch, pull, or merge.

Current Uncertainty

When a user clicks Cancel, Sourcetree sends an interrupt signal to the underlying Git process. This often leaves temporary index or lock files that a manual cleanup may need to remove. Conversely, a timeout configured in Preferences → Git triggers a Git‑level abort that typically relies on Git’s own cleanup logic, but it can prematurely terminate legitimate operations if set too low. Additionally, when both a user cancel and a timeout fire simultaneously, Sourcetree displays a generic “Operation failed” dialog without indicating the cause, complicating troubleshooting.

Specific Questions

1. After invoking the UI Cancel during a fetch, does the repository retain any lock or temporary files that require manual removal?
2. When a timeout aborts a merge, does Git automatically release all locks, leaving the repository in a consistent state?
3. In a scenario where a user clicks Cancel shortly after a timeout is about to trigger, how does Sourcetree distinguish the failure reason in the final status message?