[MSHADE-148] Don't attach exclusions that are already present#16
[MSHADE-148] Don't attach exclusions that are already present#16philipl wants to merge 1 commit into
Conversation
Although I don't fully understand the scenarios where we re-evaluate the exclusions of a dependency, I do know that if we do, we will add them again, leading to a false indication that the dependency list has changed, leading to a regeneration and re-evaluation of the list, which then continues indefinitely, as the the same exclusions are added again and again. To address the leaf problem, let's ensure that we never add an exclusion that is already present. That way, even if reevaluation occurs, it will not result in changes.
eolivelli
left a comment
eolivelli
left a comment
There was a problem hiding this comment.
What about adding this logic inside
addExclusion ?
Can you add a test case as well?
|
addExclusion is generated code in core maven. That’s also why Exclusions don’t have an equals() implementation. I’d add a test case if I could come up with a repro case that triggered on its own. I added some discussion in the JIRA. |
|
Note, that the fix in #15 looks like the fix for why exclusions get reevaluated. Even if that is merged, this fix should be taken too for correctness. |
maobaolong
left a comment
maobaolong
left a comment
There was a problem hiding this comment.
I'm +1 with this PR, hope to merge this before next release. I've patched and tested this PR, it can worked without stuck.
…n shade bug A bug in the Maven Shade Plugin may cause an infinite loop (livelock) while creating the dependency reduced pom on parallel builds It seems to happen if the added dependencies are already transitive dependencies Likely fixed by apache/maven-shade-plugin#16
|
I also have this issue when building Apache Camel with a parallel build, similar to the above example. |
…n shade bug A bug in the Maven Shade Plugin may cause an infinite loop (livelock) while creating the dependency reduced pom on parallel builds It seems to happen if the added dependencies are already transitive dependencies Likely fixed by apache/maven-shade-plugin#16
|
Resolve #505 |
1 similar comment
|
Resolve #505 |
7 participants
Although I don't fully understand the scenarios where we re-evaluate
the exclusions of a dependency, I do know that if we do, we will add
them again, leading to a false indication that the dependency list has
changed, leading to a regeneration and re-evaluation of the list,
which then continues indefinitely, as the the same exclusions are
added again and again.
To address the leaf problem, let's ensure that we never add an
exclusion that is already present. That way, even if reevaluation
occurs, it will not result in changes.