Downloading dependencies

[sinasamavati]

I don't like cloning a repository while I just need the code not its whole history, so GitHub archives come in handy. I think downloading archives (in an optional manner) is better than cloning repositories, so git, mercurial or whatever can no longer be required.

Any thoughts on this?

[blt]

If you're going to rely on just Github archives you've lost any code that's not on Github. As an end-user, I'd prefer the ability to clone a project at a given commit and, optionally, have the ability to specify a tarball URL.

[tsloughter]

Why not best of both worlds, git clone --depth 1

[blt]

Oh, didn't think of that. I like that too.

[sinasamavati]

@blt Absolutely, that's why I said in an optional manner, further, in a possible form. GitHub allows us to download a tarball at a specific commit/branch/tag, but about other hostings, making an archive builder can be useful.

@tsloughter that's pretty cool, but not efficient when a commit is specified, right?

[seth]

Using an archive seems reasonable if you are pulling a tag. Otherwise, I think you really want the SHA available so you know what you have.

So perhaps that's a +1 to git clone --depth 1?

[sinasamavati]

Aha, right.

Yes, +1 to git clone --depth but I'm not sure if it can work with a specific branch or not.

[tsloughter]

What do you mean? You can get a specific branch the same way you would before: git clone <repo> --depth 1 --branch <branch>

[sinasamavati]

Oh, yes, that's it.