Hi, On Wed, 11 Sep 2002 01:18, Robert wrote: > What I would like to know is if you actually look at the source code > from an open source project (Jakarta or otherwise) and have that > influence your usage of that particular API/Package. If not, are there > any other criteria that you use to evaluate something, such as its > following, reputation, etc.
It really depends upon what you want to use it for. I generally try to work out the cost vs benefit ratio factoring in risk. Some rules of thumb; If I start to rely on something and it fails completely then I want to know that I am capable of replacing or rewritting it. Replacement is easy if it is a standards implementation (and multiple implementations exist). Rewritting is an option for small clearly defined projects. "Complete" solutions and "one size fits all" philosophies should be avoided like the plague. I want to be able to mix n match toolkits and if this is easy then it is likely the underlying toolkits are lightly coupled and cohesive. Think of web application frameworks. Some want to give you everything from the presentation layer, through dispatch layer, through service and buisness logic layers. Some focus one aspect. ie It is easy to use Sitemesh for templating, struts for dispatching, Avalon for service layer, torque for persistence and velocity for presentation. If I decide I prefer the icky jsps I could easy swap velocity for jsp. If I decide I need to use OJB rather than torque then that should also be an easy option. Some toolkits don't allow this. Projects with recent releases are better than those without unless I monitor development lists ... and ... Projects with documentation are better than those without unless I monitor development lists. Also you will notice that opensource developers often follow a certain behaviour pattern as they become more "mature" opensource developers. ie Many start as idealistic GPL advocates who have a toolkit that will solve all your problems (that only they can maintain). Later many of these same developers move to more liberal licenses (ala LGPL, BSD, Apache) have smaller focused toolkits and find it easy to move between different projects. I prefer to use code by the more "mature" developers as has less road bumps, is more consitent and stable. The other thing I like to look at is reputation. I know certain groups/people who I think are good developers and will give whatever they do a looking at. For example, the developer of MX4J also maintains Foxtrot (a swing API for alternative style thread management). I looked at it purely because I respected the authors work on MX4J. If other people I respect also recomend X then I will look at X (thats how I ended up using Sitemesh and how I will end up using OJB). The stability and longevity of the project is also a good indicator for a project. Code quality is a fine thing to look at but does not really come in as a factor if I just want to use the project. I use quite a bit of crappy code but I don't care because I never see it and never have to maintian it (try looking through the JDK src one day). On the flipsode, sometimes really nice code (think Craigs work on Tomcat4/Catalina) is not enough. When I looked at TC4 code I really liked it but it was not as production worthy as TC3.x stream so I didn't use it for a long time. -- Cheers, Peter Donald "Man's mind stretched to a new idea never goes back to its original dimensions." -Oliver Wendell Holmes -- To unsubscribe, e-mail: <mailto:[EMAIL PROTECTED]> For additional commands, e-mail: <mailto:[EMAIL PROTECTED]>
