Hacker News Reader

Comment by discreditable | Hacker News Reader

Correct me if I'm wrong but doesn't a git clone involve downloading the project's entire git history as well as $CURRENT_VERSION? That seems a little excessive.

View on HN

Ironically git actually supports TAR for this exact thing: git archive exports tar files.

You can do a low depth to minimize that if you want (--depth=1 or something like that if I remember correctly). You could just rm -rf the root .git directory after but that is a download bloat issue.

That said this should work:

    git clone --depth=1 --branch=master git://myrepo && rm -rf myrepo/.git`

but.. yeah. Not really great.

You can use `--depth=1` to only download one revision, and `--branch` to specify an alternate branch or tag.

As far as I know though you can't grab a specific revision this way (I'd love to be shown otherwise) - which is an issue when setting up dockerfiles since you want to bake the sha hash which works into the file.

While I disagree with the authors sentiment as a whole; this issue could be solved by using a shallow clone.

By default, yes. But you can use the --depth option to limit it to a clone of the most recent state.

    git clone --depth=1

will get you just $CURRENT_VERSION

Sure, but on the other hand, downloading a tarball for every minor release means you're redownloading a lot of code. Depending on how much churn there is, you may save bandwidth in the long run by keeping a clone and updating as you go.

The article suggests that this is not a big deal anymore now that computers have more storage and networks have more bandwidth.

There are still a lot of networks that don't have enough bandwidth to make git trivial. Most home networks, it's going to turn a 5-second download into 60 seconds or more. Some networks, it's going to turn a 1-minute download into a half an hour.

And what if you find a bug, and want to regress to a previous version?

What is so cumbersome about downloading a project's history?

Well, some projects have been around for a while. "All the history" can be a LOT of history.

Actually, even for the Linux kernel, it's only a few hundred megabyte.

"Only" a few hundred megabyte is "really a lot" in many situations.

My bare Linux repository comes to 1.2GB. A git gc run got it down to 1.1GB.

Oh, true. I just checked. I think I mixed up two numbers.

Reply on news.ycombinator.com