Thanks for the confirmation, I created bug #55831. I'm assuming this problem will also rear up with our (legacy build's mis)use of <antcall>, so this is actually pretty critical for us. Is there any hope of a fix?
On Sat, Nov 30, 2013 at 11:20 AM, Stefan Bodewig <bode...@apache.org> wrote: > On 2013-11-27, Steele, Richard wrote: > > > I've run into an odd behavior that I'm assuming is a defect but I'd > > like to confirm. > > Looks like one, yes. > > > <project name="augment-test"> > > > <target name="init"> > > <path id="inner"/> > > > <path id="outside"> > > <path refid="inner"/> > > </path> > > </target> > > > <target name="update-paths"> > > <augment id="inner"> > > <fileset dir="lib"/> > > </augment> > > </target> > > > <target name="a" depends="init, update-paths"> > > <echo message="${toString:inner}" /> > > </target> > > > <target name="b" depends="init, update-paths"> > > <echo message="${toString:outside}" /> > > </target> > > > </project> > > <snip/> > > > However, if I call it with both "a" and "b" the value of the "outer" path > > is empty: > > augment is a very special task in many ways, in particular it doesn't > perform it's work in the execute method but while the representation of > the piece of XML that makes up the task is mapped to a Java object tree. > > If you execute a target twice, this translation only happens once, so > the second invocation is a NOOP. But this is an implementation detail > that users should never have to deal with, so in a way it is a defect. > > Stefan > > --------------------------------------------------------------------- > To unsubscribe, e-mail: user-unsubscr...@ant.apache.org > For additional commands, e-mail: user-h...@ant.apache.org > >