[JENKINS-69899] Do not visit EmptyExpression when transforming fields declared using @Field #457
Conversation
… setting attributes
…other than RejectedAccessException in assertRejected
| @@ -418,7 +418,7 @@ private interface Rejector { | |||
| Checker.preCheckedCast(field.getType(), value, false, false, false); | |||
| return super.onSetAttribute(invoker, receiver, attribute, value); | |||
| } else { | |||
| throw StaticWhitelist.rejectField(field); | |||
| rejector = () -> StaticWhitelist.rejectField(field); | |||
There was a problem hiding this comment.
Not related to JENKINS-69899, but fixes a minor issue in 85d16bd. See #298 (comment) for details.
I am fixing it here in case we end up backporting the JENKINS-69899 fix, since it would be nice to backport this as well. That said, the cases it fixes were already broken before 85d16bd, so it is not particularly important.
| } else { | ||
| errors.addError(x); |
There was a problem hiding this comment.
This also fixes a minor test-only issue in 85d16bd. We would also want this backported to avoid introducing broken tests.
| @@ -62,7 +62,7 @@ | |||
| <dependency> | |||
| <groupId>org.kohsuke</groupId> | |||
| <artifactId>groovy-sandbox</artifactId> | |||
| <version>1.30</version> | |||
| <version>1.31</version> | |||
There was a problem hiding this comment.
BTW I think it would be a good investment of time to inline the lib into this plugin repo.
There was a problem hiding this comment.
Yeah, enough of the pieces are in place now that it should be straightforward. I think the main issue would be checking to see if a new PCT hook is needed. I will try to do it as a Hacktoberfest project.
See JENKINS-69899 and jenkinsci/groovy-sandbox#94.
Ensure you have provided tests - that demonstrates feature works or fixes the issueTests are in groovy-sandbox