Hacker News Reader

"refs" are not behind-the-scenes things, they're the things you want to operate on. It wouldn't be like having to learn JPEG superblocks, it's more like learning what a pixel is.

You need to understand Git's data model, not its plumbing. No VCS will be useful for anything that isn't trivial if you don't understand its data model.

View on HN

I disagree. We shouldn't have to know how a branch or commit is stored. Whether git interally makes a new ref, copies everything to a new directory, or does some hashing magic doesn't matter to the user.

I agree that we end up needing to know about refs because of git's user interface, but I consider that a flaw or limitation.

It's not about "how a branch is stored". We're not talking about objects, packs, blobs or any other stuff I never actually had to learn about while using git. We're talking about refs - the primary way for users to reference commits and other things in git. Just about every command you use takes refs as arguments. It's essential to even just imagine the state of the repository in your head. It's what the first tutorial you read about git should explain to you.

Reply on news.ycombinator.com