I attended one session today on Subversion. The presentation was done by Brian Fitzpatrick who is one of the developers for Subversion. The talk was another walkthrough format where a tour of Subversion is given and some discussion on how the decisions were made in its design.
This talk was not as fulfilling for me as yesterday’s talk but that is mostly because I’ve been using Subversion pretty heavily and no fault of Brian. I didn’t really learn anything that I didn’t already know to some degree. But the talk was excellent as an introduction into Subversion.
Brian did highlight a few coming attractions for Subversion:
- Merge and branch tracking
- Operations logging
- Remote management
- Relational databases backend
- Work on 2.0 is at least in the planning stage
I’m not sure any of these features are really going to be something I want or need. The decision to implement support for relational database backends seems a bit curious to me. I’m not sure what benefit they expect users to receive. RDBM’s are notorious for handling BLOB/CLOB data so it can’t be for performance. It may be possible to scale better or to assemble history/reporting more efficiently. But there is significant overhead and I’m assuming re-architecture to integrate a RDBMS.
Brian made a reference that implied that in some ways, Subversion has done what it needed to do. I think this raises an interesting issue. In proprietary software, companies must iterate on their products in order to gain more revenue. Many commercial products are victims of this senseless feature creep. But an open source project could in theory actually be “done”. CVS got to this point. Apache 1.3 is another. Is there room in FOSS to have a project be “done” without it actively changing. I’m as guilty as anyone in assuming that projects that have little code change are dead. What is the right metric? Is there a way to recognize “completeness” with a notion of closing a project to new features?
My colleagues from SourceLabs have all arrived. I expect tomorrow to pick up considerably as the keynotes get started. It will also be the first day of the product booths. I guess it is time to go to work then.
Post a Comment