MergeLoopBlog
- Fix failing CI in the merge queue, without sending the PR back
A required check that fails in the queue usually ejects the PR. How MergeLoop's agent commits a fix on the PR instead, what it may touch, and when it gives up.
- How to merge stacked pull requests without breaking main
Stacked PRs make review easier and merging harder. How a merge queue lands a gh stack stack all at once, what happens when one layer fails, and what to set up.
- What is a merge queue, and when does a team need one?
A merge queue tests each pull request against the code it will land on, so main stays green. How one works, what it costs, and the signs you need one.
- bors-ng alternative: a maintained, label-driven merge queue
bors-ng is archived and its maintainers point to GitHub's queue. What to move to, what a migration involves, and what stays the same.
- How a merge queue batches pull requests, and finds the one that broke the build
Testing queued PRs one at a time merges a few an hour. Here is how MergeLoop tests them together in one CI run, and bisects a failed batch.
- A merge queue for private repos on GitHub Team
GitHub's merge queue needs Enterprise Cloud on private repos. What to use instead, what it costs, and what to check before you switch.
- How to resolve git rebase conflicts automatically in CI
A rebase conflict stops the queue and waits for a human. Here is the safe way to have an agent resolve it, verify it, and merge.