I am interested in a system that allows for distributed development with an authentication piece.
What do I mean by that?
Ok so lets take SVN, SVN keeps track of revisions and doesn’t care who submits, as long as you have the right to submit you can submit, really, to any part in the repository. Where does my system come into play? Being able to granulate access control and give a stackoverflow like feel to the environment.
In the system I am describing we have 4 users Bob, Alice, Dan, Joe. Bob is a project managed, Alice and Dan are programmers under Bob and Joe is a random programmer on the internet who wants to help. Ideally in this system, Bob can commit any changes and won’t require approval. Alice and Dan can commit to their branches, or a branch, but a commit to the trunk would need approval by Bob.
This is where Joe comes in, wants to help, however, you just don’t want to give him the keys to the kingdom just yet so to speak, so in my system you would setup a “low user” account. Any commits that Joe makes would need to be approved by Dan, Alice or both. However, in the system, Joe can build up “Karma” where after so many approved commits it would only need approval by one of the programmers, and then eventually no approval would be necessary.
Does that make sense and do you know if a system like that exists? Or am I just crazy to even think such a system/environment would be possible?
A question that always needs to be asked is: Why is your solution better than existing solutions?
For instance, why not use a distributed system and have random users push to someone who can then pull those changes in if necessary. Over time, if the developer proves himself, just give him access as an admin.
I know it would be cool to have all this automated, but committing source code doesn’t happen so often that an administrator couldn’t manage adding people by hand.