2 out of 3 GitHub forks are completely empty | Hacker News Reader

2 out of 3 GitHub forks are completely empty

blog.garlicsim.org

19 pointscool-RR21 comments

Just because the repository is empty doesn't mean the forkers didn't modify the code. I've forked many a project, modified it locally, and not pushed it back up for a variety of reasons not limited to: laziness, abandonment, and shameful code that somebody would likely write a scornful blog post about. ;)

I've done this too. Let's me create a github repo where "master" is fixed at a known version (to me) and I can clone it to any machine without worrying about remembering a the hash for the tag I want. I can push/pull at will and merge in changes from upstream whenever I desire. Sometimes local changes stay local, sometimes they get pushed out.

Github and Bitbucket aren't only for working on code with other people, they also let us consume projects at the rate that we want to consume them.

I usually clone the original repository. And after I make my changes, then I fork it and push to it. Most of the times there is no point to fork it before making changes.

Unfortunately, this "research" is flawed. If you pull in changes just right, the network view does not show the forks as branch off. I believe cherry-picking commits does this, and rebasing onto those commits may also do the same. Feel free to correct me, but that's been my personal experience.

So... someone could make a fork, contribute code, and then not have his name shown on the Network view? Does this happen to actual contributors?

The network view is not really intended to give a complete list of all contributors. It is intended to show the network of the different branches/forks. And it shows only a small time window (starting with the most recent time). So if there were commits earlier by other people and they were merged in, you will probably not see them.

I think under commits, the contributors list should be the requested information.

I'm pretty sure that the Network view is complete. I have around 3,000 commits over 2 years and they're all listed in my Network view:

https://github.com/cool-RR/GarlicSim/network

Thats not the same as you pulling in the commit in some way that obscures the contribution. I've done so on quite a few repositories (and I have enough repositories that all sorts of git nonsense has been tried). The big thing I try to conserve is attribution; I am not always the author of a changeset and I like to give credit where credit is due.

That said, you can also delete a commit from the Network view by issuing the command "git push -f". So no, I can't credibly say all your commits are listed, although I would hope you don't commit that unholy sin.

Yes, it has happened to me (not that I am upset). For example, the project lead can manually apply my diffs, etc.

I think some people fork because they just want a copy of it in case it goes away. That's what I've been told at least.

Last I remember, on Github, when you delete a repository any forked repositories are also deleted.

OK. I did not know that. I've never tried it either.

Really?! That seems incorrect. I don't want my code disappearing because someone deleted their repository.

Well let's test it. https://github.com/jonursenbach/fork-this

Fork it, commit something to your fork and I'll delete my repo.

challenge accepted: https://github.com/issackelly/fork-this/file-edit/

Deleted my repo. Looks like your fork is still alive and no longer marked as a fork.

confirmed

Seriously? That sounds like an extremely flawed feature.

It's only true of private repos, as of quite a while ago.

IIRC, only true of private repos.

https://github.com/blog/181-repo-deletion-for-everyone

Reply on news.ycombinator.com