Peter Donald wrote:
>
> oh. I get it .. but wouldn't you have same problem as it exists now.
> Namely the fact that taskdefs have to be registered before tasks are
> instantiated. (as in this wouldn't work with current architecture)
Given:
<javac>
<taskdef>
<foo>
The current logic would be:
new javac().init();
new taskdef().init(); <== problem!
new foo().init();
javac.execute();
taskdef.execute();
foo.execute();
Now if this were:
<compileTask>
<foo>
This would be processed as:
new compileTask().init();
javac().init();
javac().execute();
taskdef().init(); <== OK!
new foo().init();
compileTask().execute();
foo().execute();
Question for the group: this solves this particular scenario. Are there
other scenarios that this does not address which should be considered at
this time?
- Sam Ruby
