Every so often, we decide to “clean things up”. In this particular episode of organisational optimism, the goal was innocent enough: rename all branches from

Release/<branch-name> to release/<branch-name>

Lowercase. Consistent. Civilised. What could possibly go wrong?

Challenge 1: Case Sensitivity, or “Git Knows Better Than You”

The first obstacle is that we often believe that Release/1.0 and release/1.0 are two entirely different branches. As far as your local file system is concerned, they may very well be the same thing, or not.

Git, in turn, is case-sensitive. This leads to confusion, errors, and a general sense that you have personally offended your version control system.

Microsoft documents this behaviour clearly, including why Azure Repos enforces it and why case-only changes are problematic: Git case sebsitivity.

The short version: Git is right, the file system is awkward, and you are caught in the middle.

Challenge 2: There Are No Branch Folders

Despite what the slash in Release/1.0 suggests, Git does not have branch folders. There is no Release directory. There is no tree. There is no hierarchy.

The <folder> portion of a branch name is merely a naming convention. It is a label. A polite fiction. A comforting illusion we all agree to maintain so that we can pretend our repositories are tidy.

Internally, Git stores branches as references. The slash just happens to look like a folder, much like a cardboard box looks like furniture.

Challenge 3: There Is No Rename Feature

There is no rename branch button.

Not in Git. Not in Azure DevOps. Not anywhere sensible.

This is presumably because renaming a branch would be too easy, too humane, and would deprive future engineers of valuable character-building experiences. So instead, we simulate a rename by creating new branches and deleting old ones. Carefully. Repeatedly. With feeling.

The Only Way Forward

Because Git refuses to acknowledge that Release and release are “basically the same thing”, on Azure DevOps with case-sensitivity disabled, we need an intermediate step. A neutral ground. A diplomatic buffer. Let us call it Temp.

Step 1: Clone the Repository

git clone https://dev.azure.com/your-org/your-project/_git/your-repocd your-repo

Step 2: Rename Release/ to Temp/ Locally

Fetch all branches first:

git fetch --all

Rename the branch you are currently on:

git branch -m Release/1.0 Temp/1.0

Repeat this for all branches under Release/*:

git branch -m Release/1.1 Temp/1.1
git branch -m Release/2.0 Temp/2.0
...

Step 3: Push the Temp/* Branches to Azure DevOps

git push origin Temp/1.0git push origin Temp/1.1git push origin Temp/2.0Show more lines

Step 4: Delete the Release/* Branches from Azure DevOps

git push origin --delete Release/1.0
git push origin --delete Release/2.0

At this point, the repository no longer contains Release/*. Everyone takes a deep breath.

Step 5: Re-clone the Repository

Yes, really. This avoids local case-confusion and reference residue.

cd ..
rm -rf your-repo
git clone https://dev.azure.com/your-org/your-project/_git/your-repo
cd your-repo
git fetch --all

Step 6: Rename Temp/ to release/

git branch -m Temp/1.0 release/1.0
git branch -m Temp/2.0 release/2.0
...

Step 7: Push the release/* Branches to Azure DevOps

git push origin release/1.0
git push origin release/2.0
...

Step 8: Delete the Temp/* Branches from Azure DevOps

git push origin --delete Temp/1.0
git push origin --delete Temp/2.0
...

Final Thoughts

What began as a simple case change turned into a multi-stage choreography involving temporary branches, multiple pushes, deletions, and a ceremonial re-clone.

Git is powerful. Azure DevOps is correct. The process is logical.

It is also impressively unintuitive.

I sincerely hope that someone out there has a better idea.