Complex Merge Scenarios

Advanced Git

Amanda Crawford-Adamo

Software and Data Engineer

Flight travel data pipeline repo

A graph of the flight travel data pipeline repo directory and files. The git working directory has 4 four files: etl.py, load.py, transform.py, and ingest.py. The data directory has the flight_data.csv dat folder. Additionally, the git folder is also shown.

Git squash merging

Functionality

Creates a single new commit on the target branch
Combines all changes from the source branch
- Adds a regular commit with one parent (unlike recursive strategy)
Added to target branch (not feature branch)
- Doesn't preserve the detailed commit history of the source branch

Advantages

Clean and linear history
Simplifies code review for large features
Easier to revert the entire feature

Merge squash example

The data-cleanup branch diverges from the main branch after the second commit. The main branch has an additional commit. The data-cleanup has two additional commits.

Merge squash process

Checkout the main branch
```
$ git checkout main
```
Create a squash commit of all data-cleanup changes
```
$ git merge --squash data-cleanup
```

Commit the squash commit to main branch history

$ git commit -m "Implement and optimize data cleanup"

Merge squash result

A graph that shows to branches: main and data-cleanup branch. Main four commits. Feature branch diverges after main latest commit with an additional commit in it's branch. Main latest commit is a single parent commit object that contains the squashed commits of the data-cleanup additional commits.

Git octopus merge

Functionality

Merges three or more branches at once
Creates a single merge commit with multiple parents
Best used when branches don't conflict with each other

Advantages

Useful for integrating multiple independent features simultaneously
Used for synchronizing several release branches for different versions of a project

Octopus merge example

A depiction of four development branches: main, ingestion, transform, and load. The ingestion, transform, and load branch diverges from main branch after the first commit, with each having an additional commit. The main branch has one additional commit as well.

Octopus merge commands

Git Octopus Merge Command

git merge -s octopus

Example

$ git merge -s octopus ingest transform load
Trying simple merge with ingest
Trying simple merge with transform
Trying simple merge with load
Merge made by the 'octopus' strategy.
...

Octopus merge result

Summary

Squash merge

Simplifies history, combines multiple commits into one
Use squash merges for a clean, simplified history
```
git merge --squash <source_branch>
```

Octopus merge

Preserves branch structure
Merges multiple branches simultaneously

Efficiently integrating multiple parallel developments

git merge -s octopus <branch 1> <branch 2> <branch 3>

Let's practice!

Advanced Git