July 25, 2019
I have a habit of making longer git branch names. I use the branch name as a way to communicate what and why I’m working on something.
For example, beyond just storing the ticket number, I add details like
bug, the name of the section of the code, a terse summary of the problem I’m working on and then which part of that problem.
Here’s an example:
I’m working on a feature. It’s for ticket 826. It’s part of the add-edit app. The feature is adding dynamic inputs. I’m focusing on the config-file aspect.
One of the reasons I like this approach is that it lends itself to making multiple, smaller feature branches that can be viewed in isolation and/or in a cascading fashion (by changing the target of the branches).
It turns out there’s a peculiar limitation of this approach — I can’t go another level deep without changing the branch.
If I tried to create a new branch
feature/826/add-edit/dynamic-inputs/config-file/smaller-piece for example, git would throw an error along the lines of:
fatal: cannot lock ref 'refs/heads/feature/826/.../config-file/smaller-piece': 'refs/heads/feature/826/.../config-file' exists; cannot create 'refs/heads/feature/826/.../config-file/smaller-piece'
This is a known limitation of git — branches, despite my use of
/ are not directories — and git does not allow sub-branches in this way.1
Fortunately, I know how to rename a branch, so if it really mattered - I could have two branches: 2
The more you know!
Thanks for reading! My name's Stephen Weiss. I live in Chicago with my wife, Kate, and dog, Finn.
Click here to see the archives of my weeks in review and sign up yourself!