On Wed, 30 Jan 2002 21:40, Stefan Bodewig wrote: > On Wed, 30 Jan 2002, Peter Donald <[EMAIL PROTECTED]> wrote: > > I vaguely recall someone advocating that we post fix all our tasks > > with "Task" (was that you Erik?). > > Why?
Well the main reason I know of is to distinguish easily between tasks and non-task objects ;) In ant1.x the non-task objects are only adaptors and helpers for main task object. In Ant2 there will be a far broader range of types; mappers, conditions, tasks, and so forth Besides it is the pattern most tasks written outside ant follow ;) > ExecuteOn.java used to implement <execon> (something you could guess > by then) and we later renamed that task as <apply> - so people have to > look up which class implements <apply>, they cannot guess it. > > Using above rule, the class would have been ExecOnTask - how would > that improve the situation? Same for CallTarget.java that would > be CallTargetTask, not AntCallTask ;-) Not really talking about existing Ant1.x where the names are already set in stone. Talking more in context of new tasks to ant1.x and ant2. > Maybe it would be better to not place any classes into the taskdefs > package that are not implementations of tasks but utility classes? Thats another possibility. However that implies we have to use inner tasks a lot more. Doing so raises the complexity of the code and makes it harder for new developers to quickly become accustomed to the code. This is no > > any objections or not? > > Other than "I don't like it"? No. :) -- Cheers, Pete -------------------------------------------------- The fact that nobody understands you doesn't mean you're an artist. -------------------------------------------------- -- To unsubscribe, e-mail: <mailto:[EMAIL PROTECTED]> For additional commands, e-mail: <mailto:[EMAIL PROTECTED]>
