On Sat, 20 Oct 2001 11:32, Jose Alberto Fernandez wrote: > > <projectref> is different as is <import> (antlib). They both need to be > > specially handled my container because they cannot be done in task-land > > without compromising the design of the container. > > There are other ways to control the container. It does not have to be based > on hidden APIs. Just because in your design is done that way does not mean > that is the only way to do it.
err ... okay - So how would you propose an API that was not tightly coupled then ? > > > My point is that there will always be the posibility of new things > > > being declared that cannot conform with either properties nor > > > datatypes. I do not want to change core everytime one of those show up > > > (and I have said before I can think of several candidates, but that is > > > not the point). > > > > Think of several and lets evaluate them. > > I can give you several, ho hum. > but lets just mention one: > > Define tasks using a scripting language (like javascript) and allow > invoking such tasts simillar to regular tasks. This will definetly need > some new datastructure to store the definition which differs from what > regular taskdef do today. No it doesn't. > I do not care if you really like this or not, the point is that as new uses > appear we may have new things that need to be defined. To think that we > have covered all possible basis forever is unrealistic, if not arogant. Still no example yet. -- Cheers, Pete *--------------------------------* | Every rule has an exception, | | except the rule of exceptions. | *--------------------------------*
