Atom is still in Beta and occasionally demonstrates unexpected behaviors. As eager beta-testers, we as users would love to be able to provide useful feedback about how we triggered various error conditions – but the nature of the tool is that we are often interacting with it subconsciously, and cannot always recall the exact keystrokes which caused the failure.
Fortunately, computers are very good at remembering that kind of thing, and have a somewhat deeper understanding of what user interactions are doing. In this vein, I would like to propose that Atom maintain a buffer of internal call names, possibly with some arguments, which could be dumped to a buffer upon request. This would help us provide better failure reproduction cases, and should help Atom become a better editor.
For privacy concerns, this behavior could easily be made opt-in, and the buffer could be kept “reasonably” short (i.e. long enough to provide useful information for error reproduction, short enough to not contain an entire user’s interaction history).