[peripetylabs]

Perhaps I'm misunderstanding: I thought one goal of DVCS was to remove central points of failure? In that sense, isn't a central "hub" regressive?

I wonder if there's a way to host Git repositories with static files, say, on Amazon S3... That would be neat.

[mattdeboard]

>Perhaps I'm misunderstanding: I thought one goal of DVCS was to remove central points of failure? In that sense, isn't a central "hub" regressive?

This meme is getting really, really tiresome. Github being down is NOT a central point of failure. Most people know that setting up your own git server is trivial, literally a 3-4 step process. We know that we don't lose our files, our history, our working tree, etc.

The "git" in Github is easily replaced. The "hub" part has its own value. The communication tools, the well-presented diffs, the inline-editing capability, issues, wiki, etc. That's the value people are gnashing their teeth over.

[peripetylabs]

So it seems we agree. I was asking: why use Github at all? Most of the value added of Github can be done with CSS and simple network analysis.

[mattdeboard]

Because why reinvent the wheel?

[zalambar]

Git is distributed and there's no reason you should have to stop working, or committing, just because github is temporarily unavailable. At least for dependencies only on git. Losing access to wikis, pull requests, and issues may be a problem for some teams.

http://ozmm.org/posts/when_github_goes_down.html has a good summary of quick ways to keep using git without github.

[bitcartel]

Fossil has built-in issues and wiki. It just never became popular.

[threedaymonk]

The only prominent Fossil user I'm aware of is Zed Shaw, and he's since recanted after losing data and having trouble upgrading: http://sheddingbikes.com/posts/1306005291.html

[zrail]

jgit, the version that Eclipse uses, has native S3 support. I run my own local git server that transparently backs up every push to S3 this way.

http://blog.spearce.org/2008/07/using-jgit-to-publish-on-ama...