The goal of this workflow is to have a cleaner history, which makes things easier to review when you go back in time. In general, there's nothing too complex, but it includes rebasing, which if not done right and with care can cause loss of work. It's gotten a lot better and there are ways to still recover the work, so I am less worried, but the warning is still there.
One way of explaining rebases is thinking of them as .patch files. Imagine
each of the commits you are rebasing is a .patch file and you apply that patch
file one after the other on top of a new codebase. Each patch will then create a
new commit.
Normally, the .patch applies cleanly, but as you have probably encountered, it
sometimes fails to apply properly. This is when you will put on your surgeon's
cap and work carefully. It's pretty much the same as sorting out merge
conflicts, but a merge commit is more easily revertible. The other problem with
rebases is that they might require you to force-push, which is again dangerous,
but in this workflow you would only be force-pushing to the feature branch you
are working on and about to merge, so it's not a huge deal.
Tip
All the git configurations mentioned in this workflow
are available in example.gitconfig. Copy the settings
to your ~/.gitconfig or run the equivalent git config --global commands.
- Pull rebase keeping merges
- Production/Staging branches
- Pull requests/feature branches
- Rebasing
- Pull request review
- Hotfixes to production
- Optional squashing
- Commit messages
TL;DR
git pull --rebase-merges
Tip
You can configure this as your default pull behavior with
git config --global pull.rebase merges
This will fetch whatever is in the remote and reapply your local commits on top of the new code. This eliminates unnecessary remote merge commits.
The merges option keeps your local merge commits, if any. This prevents
accidentally dropping those on the main/staging branches.
Because this is just a rebase of your local commits, no force push is necessary.
Conflicts can happen, so you can either fix them, commit, and continue the
rebase (git rebase --continue) or abort it (git rebase --abort) and go back
to pulling normally if you want to be cautious: git pull --merge.
Staging should at all times be deployable to Production.
TL;DR
git checkout maingit pullgit merge staging --ff-only
Tip
You can configure this globally with
git config --global branch.main.mergeOptions --ff-only (sets the default merge
strategy for the main branch).
A common scheme is having at least one production branch (main) and a staging
branch (stage, staging, develop) that's always where code lands before
being merged onto the production branch.
Since the history between staging and production should ideally always be the
same, having a merge commit on main from staging makes no sense. To avoid
this, you should normally merge with --ff-only, which performs a fast-forward.
If this fails, it's because the history of the production branch has diverged
and needs to be fixed accordingly.
Assumes PR/feature branches are created from a staging branch named
staging.
PR/feature branches are short-lived; they must be removed once the work is merged.
TL;DR - when you are ready to merge the PR/feature branch
git checkout staginggit pullgit checkout feature/branchgit rebase staging(can cause conflicts which you'll need to fix)git push --force-with-lease --force-if-includesgit checkout staginggit merge --no-ff feature/branchgit pushgit push origin :feature/branch(removes remote branch)git branch -d feature/branch(removes local branch)
Tip
You can configure safer force-pushing as the default with
git config --global push.useForceIfIncludes true (automatically adds
--force-if-includes when --force-with-lease is used).
Note: there's currently no configuration to make --force-with-lease the
default for pushes.
Tip
You can also configure no-fast-forward merges as the default with
git config --global merge.ff false (sets --no-ff as the default merge
strategy for all branches).
Note: we've already configured main to use --ff-only when merging into it,
which is more restrictive and takes precedence for that specific branch.
If we merge PR/feature branches as-is, multiple PR/feature branches can have commits happening at different times. While this is acceptable, it gives a much clearer history graph if we rebase first. Furthermore, rebasing removes the merge commits that we might have accumulated while keeping the feature branch up to date, which are not important for the final history.
This should be done as the last step just before merging the branch into staging, which is especially important if the feature branch is being worked on by multiple developers.
You should push your rebased code to the remote PR/feature branch with
git push --force-with-lease --force-if-includes just before merging.
You can then check out your staging branch and merge your PR/feature branch with
git merge --no-ff feature/branch. The --no-ff flag creates a merge commit
for the PR/feature branch so that the history remains accessible.
Note
PR/feature branches should be short-lived, so make sure you remove both
the remote (git push origin :feature/branch) and local PR/feature branch
(git branch -d feature/branch).**
It's important to use --force-with-lease together with --force-if-includes
when force-pushing rebased branches.
--force-with-lease alone can be defeated by background auto-fetches (common in
IDEs like LazyGit, VS Code, etc.) that update your remote-tracking branch
without you realizing it. When this happens, --force-with-lease thinks you've
seen the latest remote changes and allows the force-push, potentially
overwriting others' work.
--force-if-includes adds an extra safety check: it uses your reflog to verify
you've actually integrated remote changes into your local branch before allowing
the force-push.
References:
- jesseduffield/lazygit#1668 (comment)
- jesseduffield/lazygit#1668 (comment)
- https://stackoverflow.com/questions/65837109/when-should-i-use-git-push-force-if-includes
When rebasing branches that contain merge commits (such as rebasing staging
itself), use --rebase-merges to preserve the existing merge commits from
feature branches. This is a safer rebasing approach that maintains the merge
structure rather than linearizing all commits.
git rebase --rebase-merges mainThis is particularly useful when you need to rebase a staging branch that contains multiple feature branch merges and you want to preserve that merge history in the rebased result.
Note: You can use --no-rebase-merges to explicitly disable this behavior if
needed.
Tip
You can configure this as the default rebase behavior with
git config --global rebase.rebaseMerges true (sets --rebase-merges as
the default for all rebase operations).
Here's a useful emoji code you can use for verbose code reviews: https://gist.github.com/pfleidi/4422a5cac5b04550f714f1f886d2feea
To be documented. The most important element is keeping main/staging with
exactly the same history.
I am not a fan of squashing, but when used with common sense it can be helpful. If your feature branch is full of small commits that touch very few files/lines, it makes more sense to squash them than to merge the whole history.
TL;DR
git checkout staginggit merge feature/branch --squash(there's no commit here yet, but changes are staged)git commit -m "JIRA-1234: something done"(JIRA-1234 or whatever references your PM tool of choice)git pushgit push origin :feature/branch(removes remote branch)git branch -d feature/branch(removes local branch)
Text from https://github.blog/2011-09-06-shiny-new-commit-styles/:
Always include a reference to the task (Jira, Trello, ClickUp) in the summary or the description.
If at all possible, look for integration between the PM tool and the commit.
Capitalized, short (50 chars or less) summary
More detailed explanatory text, if necessary. Wrap it to about 72 characters or
so. In some contexts, the first line is treated as the subject of an email and
the rest of the text as the body. The blank line separating the summary from the
body is critical (unless you omit the body entirely); tools like rebase can get
confused if you run the two together.
Write your commit message in the present tense: "Fix bug" and not "Fixed bug."
This convention matches up with commit messages generated by commands like `git
merge` and `git revert`.
Further paragraphs come after blank lines.
- Bullet points are okay, too
- Typically a hyphen or asterisk is used for the bullet, preceded by a single
space, with blank lines in between, but conventions vary here
- Use a hanging indent
Tip
I find 50 chars or less too short. I normally use the amazing GitSavvy Sublime Text plugin that has a sensible warning at +20 characters making 70 characters a good summary line length.
Furthermore, using conventional commits can create a very nice changelog from commit messages and also encourages you to scope your commits better.
<type>[optional scope]: <description>
[optional body]
[optional footer(s)]
From the Conventional Commits specification:
The commit contains the following structural elements, to communicate intent to the consumers of your library:
- fix: a commit of the type
fixpatches a bug in your codebase (this correlates with PATCH in Semantic Versioning).- feat: a commit of the type
featintroduces a new feature to the codebase (this correlates with MINOR in Semantic Versioning).- BREAKING CHANGE: a commit that has a footer
BREAKING CHANGE:, or appends a!after the type/scope, introduces a breaking API change (correlating with MAJOR in Semantic Versioning). A BREAKING CHANGE can be part of commits of any type.- types other than
fix:andfeat:are allowed, for example @commitlint/config-conventional (based on the Angular convention) recommendsbuild:,chore:,ci:,docs:,style:,refactor:,perf:,test:, and others.- footers other than
BREAKING CHANGE: <description>may be provided and follow a convention similar to git trailer format.Additional types are not mandated by the Conventional Commits specification, and have no implicit effect in Semantic Versioning (unless they include a BREAKING CHANGE). A scope may be provided to a commit's type, to provide additional contextual information and is contained within parentheses, e.g.,
feat(parser): add ability to parse arrays.
Other useful readings: