This is an idea I’ve been hearing recently, and I think I’m coming around on.
git checkout master git checkout -b production git branch -D master
It doesn’t have to be
production, though. It could be
prime, as long as it’s clear to you and whoever and whatever interacts with your repository that this branch is the important one.
git, when you create a new project, makes the first branch
master, but otherwise, as far as I can tell, doesn’t much care what you call your branches. When you add services like Docker Hub and such to build off of your repos, those may care if you’re committing to
master or not, but GitHub and GitLab themselves, I think, don’t care what you call your branches.
(I have suspicions about Github Pages, but I haven’t checked that out.)
We used to use master and slave to refer to hard drives. We stopped around 2003. More recently, we’ve started moving from blacklist and whitelist to blocklist and allowlist and similar terms.
(Meanwhile, I know that using gendered terms to name connectors is going to hit the same issues, and will gladly stop using this term for useful adapter hardware as soon as I know the next term.)
I have seen two responses to this. The first is, more or less, master doesn’t require servant or slave. We’re not looking to rename Masters’ Degree at this time. And, it might be compelling to you. In the specific conversation I had, the other person mentioned Tae Kwon Do teachers as master; when I took karate several lifetimes ago, we said sensei, which connotes teacher but is literally elder, or person born before another. This is not, to my mind, particularly compelling, but it might be for you.
EDITED TO ADD:
Since Git was built as an open-source alternative to Bitkeeper, I think it is safe to say that the origin of git master is rooted in master/slave terminology. Any attempts to now say master refers to a
masterrecording, or a master craftsman is a false etymology.
The other response was Sure, I get the reasons and I agree, but there’s gobs of documentation, gobs of projects and gobs of code built to assume the important branch is named master, and to change it, for every person and project and service who uses Git, would be a self-inflicted problem of Y2K proportion.
I have to assume that there will be problems.
For a new project, where I know for sure that the CI/CD and other fun won’t break, sure.
For a repo that is toy code for my Arduino or my dot files, moving from
main would be painless and trivial.
Doing the same for the repo for this blog would mean knowing, for sure, that Jekyll and GitHub Pages are happy using main instead of master, and breaking that would mean I can’t blog anymore, so I would have to test and read docs and be sure.
I am maintainer of a fairly well-used Perl modules, and while nothing stops me from changing it to main, and nothing in the build and deployment process cares about branch name, but that gets to be a history thing and plays-well-with-others thing, and I would have to think hard before doing so.
And, of course, there’s work, and the repo that holds the money-making code. Making that not work is a lifestyle-affecting decision for everyone in the organization, and making that change is not a thing that I should do alone, even if I had elevated permissions sufficient to do so. But, I can say to my boss and coworkers that this is something we should consider.
I think, in the fullness of time, that this change will become the norm. I don’t know when the flag day will be, and what the default branch name will be, and I think it’s conversations we have today that will guide those decisions.
And I will be seeing if GitHub Pages and Netlify work if you use another name for your primary branch this weekend.