| Re: [iam-dev] Question on the behavior of the builder |
Abel Muiño wrote:Yes it is. I believe DiffUtils.comparePath would check if any of the array of EStructuralFeatures in the second argument can be resolved in the first argument through EObjetct.eGet method.
I thoght DiffUtils was agnostic regarding the pom structure and it just looked at whatever path you feed it.
I also noticed EObject.eGet method works only if you are looking for EReference like Build, testResource but it won't work for EAttribute like Filters. It would throw NPE if you look for EAttribute from EObject.
Since we are just comparing if the second arguments is w/in the first argument, i suggest to use ECoreUtil.isAncestor|(EObject <http://publib.boulder.ibm.com/infocenter/wasinfo/v6r0/topic/org.eclipse.emf.doc/references/javadoc/org/eclipse/emf/ecore/EObject.html> ancestorEObject, EObject <http://publib.boulder.ibm.com/infocenter/wasinfo/v6r0/topic/org.eclipse.emf.doc/references/javadoc/org/eclipse/emf/ecore/EObject.html> eObject). I'm not that sure if it would work.
It might work for NPE but I'm not sure what's the other impact if we change some of mike's implementation for DiffUtils.
|
Anyway... do we have a test case and a fix already? (in that order, please... we should make an attempt to fix things just once :D)
On Tue, Mar 10, 2009 at 9:30 AM, Allan Ramirez <agramirez@xxxxxxxxx <mailto:agramirez@xxxxxxxxx>> wrote:<mailto:kcordero@xxxxxxxxx> <mailto:kcordero@xxxxxxxxx
Hi guys,
I took a deep look on this problem and found out the cause of NPE.
When adding a dependency, line 80 ( Object o = obj.eGet( path[i]
) ) of DiffUtils is trying to get build and filters from
dependencies and we all know that dependencies is under build and
not the other way around.
Abel Muiño wrote:
By taking a look at the DiffUtils code, it is checking the
path given in the second argument.
In the quoted example, it should be checking if any of the
modified elements are inside <build><resources>. The path-baed
search also should make sure that it is not matching a
dependency exclussion, even when the literal has the same
value, the parent is not the same.
This, in theory :)
I would recommend setting up a few test cases for the
DiffUtils reproducing the problem (a change in dependencies
makes the check for changed resources succeed) and continue
exploring from there.
Sounds good?
2009/3/5 Korina Cordero <kcordero@xxxxxxxxx<mailto:iam-dev@xxxxxxxxxxx <mailto:iam-dev@xxxxxxxxxxx>>
<mailto:kcordero@xxxxxxxxx>>>
I have a few questions about the builder, especially in
ResourceBuilderDelegate.
In investigating Bug 266034 - NPE in EMF when adding a new
dependency to the pom, I found out that builder is being called
once it detected changes in the POM.
In the method handlePomChange, there is some checking being
done
to see if Build Resources, Build Test Resources and Build
Filters
were modified.
If a dependency is added, runTest is set to true, by this
snippet:
if ( DiffUtils.isPathModified( diff.getOwnedElements(),
new EStructuralFeature[] {
PomPackage.Literals.MODEL__BUILD,
PomPackage.Literals.BUILD__RESOURCES } ) )
{
runResc = true;
}
This is because the feature ID for BUILD__RESOURCES happens
to be
the same as that of DEPENDENCY__EXCLUSIONS.
Same thing with BUILD__TEST__RESOURCES and
DEPENDENCY__OPTIONAL.
But when BUILD__FILTERS is being checked, that's when the NPE
occurs since BUILD__FILTERS has a feature ID of 13 which is
non-existent in a Dependency object.
My question on the builder is: Is it supposed to compare a
Dependency or some other modified POM object to a Build
Resource
or a Build Test Resource? Maybe it would be better if we
add some
checking first if the modified POM object is indeed a Build
Resource/Build Test Resource/Build Filter before it gets to the
checking being done in handlePomChange?
thanks,
Korina
_______________________________________________
iam-dev mailing list
iam-dev@xxxxxxxxxxx <mailto:iam-dev@xxxxxxxxxxx>
https://dev.eclipse.org/mailman/listinfo/iam-dev
-- Abel Muiño - http://ramblingabout.wordpress.com/
------------------------------------------------------------------------
_______________________________________________
iam-dev mailing list
iam-dev@xxxxxxxxxxx <mailto:iam-dev@xxxxxxxxxxx>
https://dev.eclipse.org/mailman/listinfo/iam-dev
_______________________________________________
iam-dev mailing list
iam-dev@xxxxxxxxxxx <mailto:iam-dev@xxxxxxxxxxx>
https://dev.eclipse.org/mailman/listinfo/iam-dev
--
Abel Muiño - http://ramblingabout.wordpress.com/
------------------------------------------------------------------------
_______________________________________________
iam-dev mailing list
iam-dev@xxxxxxxxxxx
https://dev.eclipse.org/mailman/listinfo/iam-dev
_______________________________________________
iam-dev mailing list
iam-dev@xxxxxxxxxxx
https://dev.eclipse.org/mailman/listinfo/iam-dev