You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
When hoff misses a webhook indicating a state change (e.g. a CI run that finished for a commit on the testing branch), PRs can just get stuck indefinitely until someone notices it and closes/reopens the PR manually.
Since webhook requests sometimes get lost in the vast voids of the internet, it would be convenient if hoff would additionally poll the GitHub API for updates, rather than relying on webhooks alone.
A good start would be polling the CI statuses of PRs that are currently being tested by hoff, since a missed webhook here can stall the queue for an indefinite amount of time. (We still would want to keep the webhook for getting prompt notifications when they work.)
Missing the webhook for a merge command is less severe, since the lack of an immediate response from the bot is already telling. A failsafe could improve UX here as well, but has less impact than one for the CI statuses. There would also be a lot more entities to poll (every open PR has the potential to receive a merge command).
The text was updated successfully, but these errors were encountered:
When hoff misses a webhook indicating a state change (e.g. a CI run that finished for a commit on the
testing
branch), PRs can just get stuck indefinitely until someone notices it and closes/reopens the PR manually.Since webhook requests sometimes get lost in the vast voids of the internet, it would be convenient if hoff would additionally poll the GitHub API for updates, rather than relying on webhooks alone.
A good start would be polling the CI statuses of PRs that are currently being tested by hoff, since a missed webhook here can stall the queue for an indefinite amount of time. (We still would want to keep the webhook for getting prompt notifications when they work.)
Missing the webhook for a merge command is less severe, since the lack of an immediate response from the bot is already telling. A failsafe could improve UX here as well, but has less impact than one for the CI statuses. There would also be a lot more entities to poll (every open PR has the potential to receive a merge command).
The text was updated successfully, but these errors were encountered: