[
http://jira.codehaus.org/browse/MNG-2377?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel#action_112912
]
Jason van Zyl commented on MNG-2377:
------------------------------------
The correct thing should be done first. The case that causes the user the least
amount of pain. If some messages fly by again and again it's an annoyance, but
having a non-functional system because some file is sitting around is not good.
This one is in the repository metadata manager so I'll take a look in here as I
was just here.
> maven-metadata not re-read after failure
> ----------------------------------------
>
> Key: MNG-2377
> URL: http://jira.codehaus.org/browse/MNG-2377
> Project: Maven 2
> Issue Type: Bug
> Components: Artifacts and Repositories
> Affects Versions: 2.0-alpha-1
> Reporter: Kenney Westerhof
> Assignee: Jason van Zyl
> Priority: Minor
> Fix For: 2.1-alpha-1
>
>
> A bug in deploy (not setting filePermissions correctly on deploy) caused the
> maven-metadata...xml and the artifacts/poms
> to be unreadable.
> When building a project with a dep on a SNAPSHOT artifact on the faulty
> remote repository, the maven-metadata can not be read.
> Maven creates a maven-metadata...xml with the following content (sample):
> {noformat}
> <?xml version="1.0" encoding="UTF-8"?><metadata>
> <groupId>com.neonics.container</groupId>
> <artifactId>bundle-factory-maven</artifactId>
> <version>1.0-SNAPSHOT</version>
> </metadata>
> {noformat}
> After fixing the permissions I restart the build, but the metadata file is
> not re-read, resulting in Maven thinking
> 1.0-SNAPSHOT is the actual version and trying to get
> ...artifact/1.0-SNAPSHOT/artifact-1.0-SNAPSHOT.pom
> which does not exist.
> Touching the files on the remote repo doesn't help either.
> Only when running with -U the metadata file is retrieved (another solution is
> to purge files from
> the local repository).
> I think Maven should not create the metadata file if it can't be retrieved
> AND if the non-timestamped
> SNAPSHOT version doesn't exist.
> I can see why the placeholder metadata files reduces future requests for this
> file,
> but that's only useful if there _are_ future requests. There are indeed
> repeated requests for the pom
> and artifact, even though they also don't exist. I think that only when the
> artifact 1.0-SNAPSHOT exists
> and the metadata file doesn't, it's safe to put a placeholder metadata file
> in the local repository.
--
This message is automatically generated by JIRA.
-
If you think it was sent incorrectly contact one of the administrators:
http://jira.codehaus.org/secure/Administrators.jspa
-
For more information on JIRA, see: http://www.atlassian.com/software/jira
