Jeff - Thanks for the data on users and lines. Put a linear or even logarithmic curve on that and you will see that even though the number of committers grew the number of lines are falling [1]. It would be interesting to see who these new committers are and where they are coming from. My guess is from universities!
What I learn from the data is, that OpenACS is becoming less attractive to developers with each day. In the past a bunch of active developers committed alot. Now many of those are leaving or less active and are accompanied by many who commit less.
I believe that "marketing developers" is very important. If the toolkit is attractive then developers are attracted to implement projects they are offered with this toolkit. If it is not they will choose another toolkit. So either the toolkit is sexy or OpenACS will decline. Marketing developers is from my understanding to offer them a great toolkit that is superior enough to compete with mainstream technologies. So what we need to offer is what has been said before:
- Quality and release management. OpenACS is almost great on that. I hope that .LRN will improve as well. We have problems with regular releases, communication of the release status, upgrade scripts.
- Installation. Windows is very important. I know some folks prefer linux over windows. But windows has a huge market. Noone really questions Oracle versus Postgres because Oracle has its market. We should find a way to support windows and linux without any patches and long installlation procedures.
- Documentation. A developer is attracted by a good documentation that shows him how easily he can implement a package and where to find the API. We need a well documented and consistent API. I know the gurus now say "documentation is for the weak". This doesn't help. It's like going to the Dr. (OpenACS) and ask for medication and get the response "darwin's law baby...be strong or die". No! We need a documentation that allows a user to write his own package in 30 min. As a newbie I find it extremely hard to understand how I can define objects in OpenACS. Maybe OAK CASE [2] or the upcoming package builder will close that gap. The templating system is great but not completely documented I guess.
Anyway. It's interesting that this discussion is going on now for at least 2 years. Which is a sign to me that business opportunities for developers with OpenACS and .LRN is decreasing. :(
Ideas?
[1] http://madura.bwl.uni-mannheim.de:8080/data.jpg
[2] http://www.weg.ee.usyd.edu.au/people/laetitia/
