Steering Subversion
When the Subversion project first launched, it was blessed to have something that many much older projects still lack: a clear direction. It would have been easy — and certainly the temptation existed — to try to make Subversion all things to all people. Why settle for "a compelling replacement for CVS" (which was the goal at the time) when we could shoot for "best-of-class version control"? "Next-generation SCM solution"? "All that and a bag of chips?" Here's why: so the developers could get anything accomplished at all. It's as the old Proverb says (though admittedly in a different context), "Where there is no vision, the people perish." Without a single, clear sense of direction, you're left either meandering about with no sense of direction at all or (worse) stalemated by N senses of opposing directions. So how do the leaders of an open source software project choose a direction and set goals? And where do you, the reader, fit into all of this? Read on.
As I said before, Subversion originally had a vision of being a compelling replacement for CVS and nothing more. And that's exactly what Subversion 1.0 was. But where do you go when you've been operating with a single goal in mind and then you reach that goal? Well, in Subversion's case, the 1.0 target was never even considered as the last station on the line. The developers always knew we wanted to push well past being a mere replacement for CVS. Really, the sky was the limit on where we could have driven Subversion. The project had already amassed a feature wishlist from its own developers anxious to scratch their version control itches and from early-adopting Subversion users. And new ideas were popping up daily. So Subversion started picking away at those features which seemed to its developers to provide the most bang for the buck.
Meanwhile, our little software tool was finding its way into the hands of a different class of user than we tended to interact with in the open source arena. CollabNet packaged Subversion 1.0 into its CollabNet Enterprise Edition software, and suddenly Subversion was being adopted officially by segments of some rather large corporate enterprises. Instead of being used to replace only CVS, Subversion was displacing the likes of Microsoft's Visual SourceSafe, Perforce, and in some cases, threatening even heavyweights like IBM Rational's ClearCase. Those kinds of installations brought with them a whole new class of feature and enhancement requests.
CollabNet — as the vendor which founded Subversion, remained the greatest source of funding for its development, and delivered Subversion to the Enterprise — recognizes the value of clear goals. And it also recognizes that one of the key benefits it can provide to the Subversion open source community is the valuable feedback of its customers, which tend to shy away from public discussion forums. CollabNet has over time helped the Subversion developers to get a better sense of what its customers are looking for in an enterprise-class version control system, mostly notably with its activities around the requirements gathering, design, development, and beta testing program for the merge tracking feature recently released in Subversion 1.5.
Those of you familiar with the Subversion 1.5 release and the merge tracking development process know that this was a beast of a feature. Hard to define, hard to design, and hard to implement, this flagship feature of the 1.5 release was somewhat all-consuming at times. And the feature still isn't completely finished. But as the Subversion developers come up for air in the wake of the 1.5 release, we find ourselves at that familiar place again — the place where another major milestone is behind us and we're looking forward to the future. As always, there's no shortage of wishlist items to consider, and development is already well under way on some of those things. But we continue to learn more about our users, about their needs, about ourselves, and about what Subversion can and cannot do today. And that's where you fit in.
While it is true that sometimes programmers wish they could turn off all the noise around them so they can focus on the task at hand, Subversion's developers have always understood that there's usually some signal in that noise, a thread of meaningful input that's worth attending to. Your participation in the Subversion development community is vital to the continued success of this software. We need and want to hear what you have to say about Subversion and how it does or doesn't meet your version control needs.
To that end, allow me to make you aware of a special opportunity you will have to do exactly this. The second annual SubConf Subversion user conference is being held October 14-16, 2008, in Munich. CollabNet has worked with the conference organizers to arrange for a special roundtable discussion session (to be held Tuesday evening, the 14th) in which you, the Subversion user, can meet some of the Subversion committers in person and share your experiences with us. This is not a lecture-style session where we do all the talking. We want to listen. This is a great way for those of you averse to communicating in open source forums to make your opinions and desires known. I'll be chairing the roundtable session, and many of my colleagues will be present, too. We hope to see you there!
For more information on the SubConf conference, visit http://www.subconf.conf/ [German].
For information about the SubConf Subversion Roundtable, visit http://www.subconf.com/roundtable/ [German].
ShareThis
Unfortunately, I can't come to your conference, so I'll say it here:
The 1.5 release is extremely good. Keep up the amazing work. Can I have one feature please?
Local Branches
I don't think we need a fully distributed version of Subversion. In most corporations, distributed version control makes little sense. We just need 90% of the bang for 10% of the buck. The one nifty feature of distributed version control is the ability to *quickly* make local branches that we can revert or switch to.
Now that we have half-decent merge tracking, the pain of having these won't outweigh the usefulness of them.
This would give us 90% of the bang for 10% of the buck, so to speak.
Simon
Simon Johnson | August 27, 2008 at 02:06 PM
Thanks very much for Subversion 1.5. I can't be at Munich. How about setting up a survey of possible features and let people vote?
A feature I would like is the ability to manage auto-props on the server side.
David
David Aldrich | August 28, 2008 at 01:09 AM