[
https://issues.apache.org/jira/browse/MNG-6415?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16484595#comment-16484595
]
Robert Scholte commented on MNG-6415:
-------------------------------------
With Maven the dependency resolution defines the order, there's no such thing
as wildcards, the order is always predictable. Depending on the way the
application is created it is possible to respect the order. With MNG-6025 that
order changed.
I do understand the claim from Thomas, especially when referring to a directory
instead of explicit jars, if you know that
[File.html#listFiles()|https://docs.oracle.com/javase/7/docs/api/java/io/File.html#listFiles()]
says
{quote}There is no guarantee that the name strings in the resulting array will
appear in any specific order; they are not, in particular, guaranteed to appear
in alphabetical order.{quote}
Maven is a different story, it should not change the order if applications
relies on that (on purpose and on their own risk). Otherwise we should add an
option to randomize the order of the classpath.
> Project Artifacts Cache does not retain the order of classpath entries.
> -----------------------------------------------------------------------
>
> Key: MNG-6415
> URL: https://issues.apache.org/jira/browse/MNG-6415
> Project: Maven
> Issue Type: Bug
> Components: core
> Affects Versions: 3.5.2
> Environment: Windows 7, JDK8u144
> Reporter: Seckin Onur SELAMET
> Priority: Major
> Labels: CLASSPATH
> Attachments:
> [MNG-6415]_Fixes_Project_Artifact_Cache_classpath_order_retaining_issue_.patch
>
>
> Project artifacts cache does not retain the order of classpath entries.
> Wrong Object type used in implementation. HashSet can not guarantee the order
> of elements.
> In runtime ProjectArtifacts passed as LinkedHashSet already which is safe.
>
> Possible fix is provided in comments section.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
