On 2020-05-04 at 17:20:33, Simon Pieters wrote: > "master" is an offensive term, as it can be interpreted as being > slavery-origin terminology. See > https://en.wikipedia.org/wiki/Master/slave_(technology)#Terminology_concerns > > The Python programming language, and various other projects, have > taken a stance and moved away from offensive terminology including > "master". See https://bugs.python.org/issue34605 > > When different projects using git decide to move away from "master" as > the name of their main branch, inconsistency ensues between projects. > See https://github.com/desktop/desktop/issues/6478 (and "Related > Issues and Projects"). > > To avoid offensive terminology and to avoid further inconsistency, I > think git should use a different branch name than "master" when > initiating a repo. I don't have a strong opinion, but I like "main" > since it shares the first two characters and it's shorter. I've been busy and haven't had much time to respond to this, but I've gotten some feedback from other people on this issue and so I'll share a few thoughts. Others have pointed out that "master" meaning a canonical source may not share the problematic origins mentioned above. From feedback I've received, I get the impression that "master", even though from a different origin, brings the idea of human bondage and suffering to mind for a non-trivial number of people, which of course was not the intention and is undesirable. I suspect if we were making the decision today, we'd pick another name, since that's not what we want people to think of when they use Git. Clearly we have compatibility concerns to consider though, so if we decided to make a change, we'd probably want to make it in a 3.0, which as far as I'm aware hasn't been discussed yet. I also wondered what such a change would involve, so I did some research. It appears that if we made the obvious one-line change to builtin/init-db.c, we'd have 304 tests that fail, which is about a third of our test suite. I haven't examined any of these tests, so I don't know what would be involved in changing them. I imagine a project to do so would involve setting an environment variable in the test setup code (e.g., MAIN_BRANCH) and replacing instances of "master" with that until everything works with an alternate value of that variable. Picking the new name itself could be deferred until later, and we could choose from some popular alternatives. There's also the documentation, which at first glance seems mostly to be examples, many of which could be changed to any suitable branch name. There are a large number of those cases and someone would have to audit them all. So it looks like this would be a reasonable amount of work for someone if they decided to pick it up as a project. Since I have limited free time and am working on the SHA-256 transition, I won't be doing this, but if someone did pick it up, I would be happy to do some reviews, provide feedback, and include a few patches while doing other work in the area. I realize there isn't agreement on a direction forward or whether this is worth doing at all, but since Git usually operates by providing feedback on an initial set of patches, I thought I'd sketch out what that might look like for folks who were interested. I should point out that it's also possible for users who dislike the current name to use a template to change the default branch name like so (using the proposed "main"): mkdir -p ~/Templates/git cp -a /usr/share/git-core/templates/* ~/Templates/git echo 'ref: refs/heads/main' > ~/Templates/git/HEAD git config --global init.templateDir ~/Templates/git Then "git init" will set your default branch to "main" instead of "master". This does have the weirdness that it claims it's reinitializing your repository, but otherwise appears to work. That is of course orthogonal to changing Git itself, but is an option for folks who'd like to make a change now. -- brian m. carlson: Houston, Texas, US OpenPGP: https://keybase.io/bk2204
Attachment:
signature.asc
Description: PGP signature
