Bardi Harborow
Bardi Harborow
After almost six years, [GitHub has added its own functionality to sync forks with upstream](https://docs.github.com/en/pull-requests/collaborating-with-pull-requests/working-with-forks/syncing-a-fork). That update makes this project redundant, so I will be wrapping up and shutting down...
We should use [octokit/octokit.js](https://github.com/octokit/octokit.js) rather than manually calling the GitHub API.
We should code in something more developer friendly and compile as needed.
We should reprompt for authentication if the token expires or is revoked.
Following on from #4, we want to be able to change our minds about the level of access after we've already chosen a level. This will make #16 less problematic.
Upriver works internally by fetching upstream's sha1 hash, and resetting your forks state to match. GitHub automatically bails if doing so would lose data, but we have a "force" option...
Heroku does not provide enough "dyno hours" to run the gatekeeper instance 24/7. This will become a problem as our popularity grows.
Building on #56, this PR adds a `UseSecondaryDB` context manager which unpins within a block. (You can just review the second commit, apparently I can't stack PRs from a fork....