Did you ever wonder why the gradle-wrapper jar went from 10k in version 0.5 to
1,140k in version 0.6. It's because the jar task has been broken :(
The convention resouceCollection is always added to the jar even if you specify
another fileset. This results in strange behavior, such as:
task wrapperJar(type: Jar) {
appendix = 'wrapper'
fileSet(dir: classesDir) {
include 'org/gradle/wrapper/'
}
}
For the previous task, it will end up with the fileset from the convention
(fileSet(dir:classesDir)) in addition to the one you specified above. This
means that files which meet both specs will actually end up in the jar twice! I
didn't even know you could do that, but you can see the duplicates of
Wrapper.class in gradle-wrapper-0.6.jar (and later).
I think the problem is introduced in AbstractArchiveTask.resourceCollections
(AbstractArchiveTask.java line 229):
resourceCollections = GUtil.chooseCollection(resourceCollections,
getResourceCollections())
If you are setting up a new resourceCollections, you probably don't need to add
the convention one first.
I think the following may fix it, but I haven't been around the archive code
enough to know if there are other problems with this approach.
Index: src/main/groovy/org/gradle/api/tasks/bundling/AbstractArchiveTask.groovy
===================================================================
--- src/main/groovy/org/gradle/api/tasks/bundling/AbstractArchiveTask.groovy
(revision 1649)
+++ src/main/groovy/org/gradle/api/tasks/bundling/AbstractArchiveTask.groovy
(working copy)
@@ -226,7 +226,9 @@
}
public AbstractArchiveTask resourceCollections(Object ... elements) {
- resourceCollections = GUtil.chooseCollection(resourceCollections,
getResourceCollections())
+ if (resourceCollections == null) {
+ resourceCollections = new ArrayList();
+ }
GUtil.flatten(Arrays.asList(elements), resourceCollections);
return this;
}
I think this would be a good candidate for a new unit test!
You will need to release a new gradle wrapper with a fix for this before you
build 0.7.
--
Steve Appling
Automated Logic Research Team
---------------------------------------------------------------------
To unsubscribe from this list, please visit:
http://xircles.codehaus.org/manage_email
