Confused by what you are asking. please see my questions below. > Nicola Ken Barozzi wrote: >> >> Dean Hiller wrote, On 26/08/2003 13.59: >> ... >> >>> Currently at work we prevent this same thing by having separate source >>> trees and people are complaining up a storm. >> >> ... >> >>> This helps with that problem of allowing one to keep one source tree >>> and put up walls that prevent other packages to be used from certain >>> packages that are supposed to be independent. >> >> >> I like it. :-) >> >> I have never liked keeping separate directories for things, even if I do >> it for the benefit it has in build separation. >> > > I like a good layout of project.model. and project.view. trees; any > instance of .view packages in the model class is a complete failure of > separation and easily caught. But having more rigorous meta-rules is > interesting.
Yeah, so I thought, until a developer once put a dependency in that caused me a huge headache to get it back out. > Would it need to be a compile time process, or something you can do just > by processing the source or even the class files, a la <depends>? I wrote this task in a few hours. (Testing took a little longer). I didn't think of using the depends and going that route. That might have been a good idea. I would have had to do alot of comparisons. That is something I didn't have to do with the route I did take which was just using javac on the different packages. javac naturally failed the build when it couldn't compile due to bad dependencies. Is there a process for acceptance of a new task into ant? How does this work? Still need to fix my bug and write a test case, but after that??? thanks for any input here. dean > > > --------------------------------------------------------------------- > To unsubscribe, e-mail: [EMAIL PROTECTED] > For additional commands, e-mail: [EMAIL PROTECTED] > > --------------------------------------------------------------------- To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
