On Tue, Jul 15, 2008 at 6:39 PM, Les Hazlewood <[EMAIL PROTECTED]> wrote: > Why would you or anyone else on this list care if it might be > over-engineering, especially when I've been able to whittle away to a > minimalist solution with NO CL issues with an end result that will > require extremely little or more likely NO maintenance?
I can't speak for the others on this list (although they seem to be doing pretty good on their own :-). In my case it's simply because I think most users will find the magic fallback solution more, not less, confusing the an explicit dependency on SLF4J. Especially since that's what they are used to from pretty much every other open source project these days, be that JCL or SLF4J. [snip] > A huge reason why I'm involved in JSecurity and OS in general is > because I _can_ implement things in the cleanest possible way, they > way they should be done - the way software fundamentals should be > employed. I do it because these principals are not always possible in > the commercial world, where schedules and costs often dictate (crap) > results. This project is my breath of fresh air. I think we all feel similar about our pet projects, I know I do. However, that doesn't mean everyone agrees on principles and therefore what implementation should be chosen. Often that will lead to the need for compromising on ones principles, hopefully for the better result in the long run. In this case, there are some pretty experienced people (I'm not one of them) arguing here, that's good input for any developer. > My reduced solution works, it enables a cleaner, more flexible > deployment scenario than using SLF4J natively, and there are no CL > issues. There is nothing to learn. Its just an elegant solution. > Why some feel it is not worth trying before 1.0 status just baffles > me. There is lot to learn, every user of JSecurity will need to understand how its logging works, or they will be very confused when it falls back to a logger they had no idea was in use. Especially when the project declares a dependency on a logging facade. /niklas
