> From: Ken Wood [mailto:[EMAIL PROTECTED] > > > On a related note, while I'm thrilled to see more and > more people using Ant, I'm worried that we keep > adding subtle variations of the same thing just > to meet a specific need. Look at the number > of variations of "available" that have been proposed > in recent weeks, or the variety of logger issues, as > just two examples. > > Or, someone submits a patch to add functionality that > is already there, they just didn't know it. > > I believe the committers are doing a good job of > screening out duplicate work, but I still fear that > if we, as a group, aren't careful, we'll end up > with a 10,000 lb gorilla that can do the same thing > 30 different ways, but can do anything well or > efficiently. Ant started out as a lean, mean, building > machine, and I hope we keep it that way rather than > burdening it with every subtle variation under the > sun. Some things belong IN ant, and some things belong > in the surrounding infrastructure in which ant is used. > We have to be careful to keep the line between the two > clean. >
I agree with you here. I think the solution to this is two fold: 1) Make even easier for people to define its own task sets. The changes to <taskdef> are a wonderful start. Once we define a way of packaging multiple related tasks in one unit, I think it will be perfect. 2) Provide a library where people can contribute and share their own task libraries. If (2) where in place, then ANT the project could get out of the bussines of maintaing optional/contributed tasks and it could become the lean project it is supposed to be. We could have a voting mechanism to decide which and when this contributed tasks become part of core. The point is that people should be able to write new tasks easily by extending tasks on core. Jose Alberto
