On Jul 12, 2009, at 9:37 AM, Russel Winder wrote:
On Fri, 2009-07-10 at 16:01 -0400, Steve Appling wrote:
I have a git repo with a new Copy task that does not use Ant. It
allows us to
determine if anything was really copied (getDidWork() is accurate)
and has some
additional features from the current copy task.
Apologies if this has been debated, if so feel free to point me at the
thread with the decision.
As I understand it, Gradle will always give access to Ant and so will
always carry a dependency on Ant.
If you use Gradle the classic way, yes. But we want to make Gradle
very modular in the future. For some embedded use cases it is
important to have as less dependencies as possible. But the main
motivation for the new copy task is not driven by this (see below).
Ant has a copy function, so why
replicate this behaviour?
The ant.copy task does not provide an API as rich as we would like it
to be. A Gradle copy object should be able to tell for example if it
actually has copied anything (very important for optimizations). The
Ant tasks have not been designed to be used as an API. You just
configure and execute them. The possibility to communicate in the
other direction does mostly not exist or is very weak. Therefore we
are also working on our own native test runners as a replacement for
the Ant tasks we are using at the moment.
- Hans
--
Hans Dockter
Gradle Project Manager
http://www.gradle.org
---------------------------------------------------------------------
To unsubscribe from this list, please visit:
http://xircles.codehaus.org/manage_email
