Will Open Source Dominate Development Management?

[article]

In his "CM: the Next Generation" series, Joe Farah gives us a glimpse into the trends that CM experts will need to tackle and master based upon industry trends and future technology challenges.

Summary:
Open Source is a great concept, thmough any resisted at first. Can you remember a time when it was valid to use the argument that, "Nobody will commit to supporting it, so you're taking your chances"? Today, I see it differently. There are loads of support for open source products, even better than for commercial products in many cases. So does that make it the winner?

I don't really think so. It is a winner. We like to look at various open source tools and use them fairly heavily. But there are some aspects of open source development that I would like to consider from a management perspective. They deal with innovation and excellence; architecture and entrepreneurship.

If I'm looking for software that's going to meet a standard, I can't go wrong with open source. Maybe today it falls short, but as the momentum for a particular product development picks up, it will exceed competitive solutions. In the early 90s it seemed to be a risk to go with the GNU C/C++ compilers. It quickly became apparent, though, that Gnu would set the standard for C/C++, for Make, etc. We hedged our bets and eventually went over to GNU.

Then we started looking at other things. What about Open Office? Great compatibility, but perhaps it could use some downsizing. Not to worry, right? It's open, we could do the downsizing ourselves if necessary. Well, no. Why? I find it's almost as big an investment to adapt open source code, as it is to write a separate application. Then there's trying to get the adaptations into the core of the development stream.  Put the ideas out there in the form of feature requests and they will get picked up and implemented.

So what's the problem, you ask? Well, software is getting big and hard to manage. With so many hands in the pie, even if there's common ground on what should be part of the product, the architecture begins to suffer. There's no room for simplicity. There's little room for the orthoganal architecture that optimizes the results. Then there's momentum: just try to change the direction. The problem is that open source development is primarily a feature-driven exercise.

I know a lot of commercial endeavors are the same. In fact, Microsoft didn't excel because of its technology. There were better operating systems around, and even today I find it incredible that I can't do so many things that 30 years ago I could do with Unix, such as redirecting my display for a given application or resizing my "shell" window. So, commercial's not always better.  But it could be.

With open source you may have a visionary, and even a vision, but it's hard to attain. With a focused drive to attain a vision, a commercial development effort can yield terrific results. That's how VNC came about. That's even how Unix came about. And it's great that open source was there to pick these both up when they were in danger of losing a marketing battle. But where has the little Unix kernel gone. Because there are Open tools, widgets, gadgets, you-name-it, they have been absorbed, not because of their simple design architecture, but because of their function, and perhaps their functional architecture. As a result, my Gigabyte-memory computer takes longer to boot up than did my 1MB Atari ST of 20 years ago. I could do word processing, desktop publishing, spreadsheets, games, etc. The network wasn't there yet, so it may be a bit unfair to compare.

So my first complaint is that it's hard to take a vision forward within an open source development framework.

Now let's get to the CM side of things. Open source means CVS. There are a lot of add-ons, as well, but the open source architecture will prevent CVS from really going places. Perhaps a next generation open source CM product will appear, though!  So what's the difference between CVS and CM+, Accurev, MKS, etc.? CVS has established a framework that really won't be challenged by its development team. CM+ made some dramatic advances with allowing dedicated teams for dedicated periods of time and augment the vision.

Process workflow engine, dedicated effort.  GUI generation engine, dedicated effort. Active workspace, dedicated effort. These are much harder to do in an open source community. You have to have a vision, battle it out with other visions, and sell it to the gatekeepers, etc., all across a global network of "owners". It's just easier to take many little steps. Put 100 new features on next year's model car, though, and you don't get a winner. You have to have the vision for where the car is going, work closely with others as to how to get there and then let the feature set fall out of this work. You can't just paste the features together to get a new model.

That's why open source, though great, will remain architecturally limited. That is, unless you have big players ready to put in big investments which are then turned over to the open source community: IBM and Linux. Sun seems to be going the same route with Solaris and it will likely make a big splash at some point, but look at database technology and you will see that it’s stagnant. Who's going to make a breakthrough and open source it? It's not likely to happen any time soon because the investment will be great. Why? Because SQL is entrenched in so many other technologies. Eventually, though, we will see it and the complexities of mapping from real-world to relational and back to real-world will disappear. This is not going to be an open source effort and it will take some time.

Because of these changes, development tools that are integrated on top of SQL will abound. They will be stuck in the mud, but maybe administration and performance will improve. Still, there will always be the need for the data experts to make the application work as long as there is a purely relational model. There will be some ALM tool integration, but it will be limited, especially for demanding applications. There will also be customization, but along with it, there will be loads of scripting to manage, and the complexity will build.
 
Good management, process and data tools all working together, or even integrated into one tool, require careful architectural design. Today, open source is open source. Will open design be next? I don't really think so. Design and architectural breakthroughs are more the domain of entrepreneurs. I wouldn't be caught dead trying to develop a new fancy editor, as there's not enough new there to preclude it from being done on the open source market. It's precisely because CM is complex, and requires good architecture and design, not just a few fancy ideas, that there will always be a commercial CM tool demand.

About the author

CMCrossroads is a TechWell community.

Through conferences, training, consulting, and online resources, TechWell helps you develop and deliver great software every day.