On Wed, 31 Oct 2001 19:39, Stefan Bodewig wrote: > Seems we have three options left that didn't receive any -1s: > > (1) All tasks (and types) can be placed into an <init> construct which > is much like a target, but cannot have dependencies of its own. All > targets depend on this one implicitly. No tasks or types can be > siblings of target. > > (2) All tasks (and types) can be siblings of targets. > > (3) Identify those tasks or types that can be siblings of targets by a > flag inside the task library. > > I'm just trying to boil down the options, maybe we can finally come to > a conclusion. > > Stefan > > who prefers option (1) but wouldn't veto the alternatives either.
I would prefer (1) if we made it a normal target. Failing that (2) sounds like an option ;) I still think things that effect the processing of build file (import statements and aspect uses) should be at the top level as that is the only way we can make ant easy to pre-process/validate the build file. -- Cheers, Pete *------------------------------------------------* | The student who is never required to do what | | he cannot do never does what he can do. | | - John Stuart Mill | *------------------------------------------------* -- To unsubscribe, e-mail: <mailto:[EMAIL PROTECTED]> For additional commands, e-mail: <mailto:[EMAIL PROTECTED]>
