Right now, I’m sitting in Greg Stein’s A New Google Service for the Open Source Community presentation at OSCON where he has just announced project hosting on Google Code starting today for open source projects. This service is similar to Sourceforge, but it’s done the Google way. Here are my quick notes from the presentation:
- Simplicity, scalability, reliability
- Rebuilt Subversion on top of Bigtable (instead of using Berkley DB or filesystem)
- Complete re-think of issue tracking (simple system with labels on each of the issues, using the Google search infrastructure to easily search across the labels, titles, and descriptions without the need for complex workflows to track issues)
It’s located at: https://code.google.com/
There’s no project approval process; project goes live right away. They are working with existing projects right now to ensure that project names are not all grabbed up. They will be running a number of different analyses on project patterns to ensure that tons of bogus projects are not created. There are lifetime limits of project creation. They don’t use a CAPTCHA when creating your first project in a day, but if you try to create more than one project a day, you will receive a CAPTCHA on all subsequent project creation pages.
Sample project located at: http://code.google.com/p/hostingdemo/
There will be emphasis on e-mail messenging for issue tracking, but RSS feeds will also be available for issue tracking as an alternative. Projects can have links to Google Groups, blogs, mailing lists, etc., all of which will be reflected on the project summary page. The Subversion repository can be viewed through the Web browser.
An Open Source license must be selected when created a project, and Google is limiting this to a short list of seven licenses because Google doesn’t want to encourage license proliferation. These seven licenses are: Apache, Artistic+GPL, Mozilla, MIT, New BSD, GPL2, LGPL. (Google’s code for this service will not be available on their own service, nor will it be available through an open license.)
They will be working on how to best integrate the service with other Google services. They will be working on APIs to access and/or dump the data. However, Greg is hesitant about opening up an API because it sort of locks them in. They may think about ways to connect with Sourceforge, but, again, this means they need an API.
The service is still (surprise!) in beta.