Is there a need for a `checked::reset()` function?

[ddsjoberg]

I ran checked::check_rev_deps() twice, saving the output to the same folder. The package I was testing had been updated between runs. But the second run finished in less than 1 second. I suspect it found all the installed packages, and thought it didn't need to re-install. This is probably GREAT for finishing long jobs that may have stopped for whatever reason. But for re-testing, not as great

Perhaps it would be helpful to have a check that the output folder contains a complete check or not. If you try to run the function and save the results in a "complete" folder, you need to first reset the folder (that is, delete the contents)?

[dgkf-roche]

Thanks @ddsjoberg!

The intended use case is to re-use prior run information, especially in cases where the process was prematurely interrupted. However, I see how that can be unintuitive. Since rev dep checks are quite time intensive, we wanted to default to keeping old run information. That said, if it's unintuitive, we want to find a way for the output to communicate what is going on more clearly and what can be done to change it.

On our call we proposed defaulting to emitting a prompt to select this behavior during interactive execution by default and exposing a global option to configure the default behavior.