Re: Repository scanning problem in 1.0?
Brett Porter wrote: On 30/11/2007, ArneD [EMAIL PROTECTED] wrote: Do you think the repository scanning problem as such, as described in my original mail, should be filed as a second issue? Yes, it looks that way. - Brett I filed http://jira.codehaus.org/browse/MRM-612 In my opinion, this should have a high prio - Archiva should always be able to synchronize itself to the filesystem contents, shouldn't it. Arne -- View this message in context: http://www.nabble.com/Repository-scanning-problem-in-1.0--tf4897121.html#a14134939 Sent from the archiva-users mailing list archive at Nabble.com.
Re: Repository scanning problem in 1.0?
I don't know if this is related, but now I see a NPE in the logs when using Update Database Now: jvm 1| 2007-11-29 14:16:03,340 [Thread-6] ERROR org.codehaus.plexus.taskqueue.execution.TaskQueueExecutor:database-u pdate - Error executing task jvm 1| edu.emory.mathcs.backport.java.util.concurrent.ExecutionException: java.lang.NullPointerException jvm 1| at edu.emory.mathcs.backport.java.util.concurrent.FutureTask.getResult(FutureTask.java:299) jvm 1| at edu.emory.mathcs.backport.java.util.concurrent.FutureTask.get(FutureTask.java:118) jvm 1| at org.codehaus.plexus.taskqueue.execution.ThreadedTaskQueueExecutor$ExecutorRunnable.waitForTask(Thread edTaskQueueExecutor.java:159) jvm 1| at org.codehaus.plexus.taskqueue.execution.ThreadedTaskQueueExecutor$ExecutorRunnable.run(ThreadedTaskQu eueExecutor.java:127) jvm 1| Caused by: jvm 1| java.lang.NullPointerException jvm 1| at org.apache.maven.archiva.database.updater.ProcessArchivaArtifactClosure.execute(ProcessArchivaArtifac tClosure.java:56) jvm 1| at org.apache.commons.collections.CollectionUtils.forAllDo(CollectionUtils.java:388) jvm 1| at org.apache.maven.archiva.database.updater.JdoDatabaseUpdater.updateProcessed(JdoDatabaseUpdater.java: 170) jvm 1| at org.apache.maven.archiva.database.updater.JdoDatabaseUpdater.updateAllProcessed(JdoDatabaseUpdater.ja va:111) jvm 1| at org.apache.maven.archiva.scheduled.executors.ArchivaDatabaseUpdateTaskExecutor.executeTask(ArchivaDat abaseUpdateTaskExecutor.java:78) jvm 1| at org.codehaus.plexus.taskqueue.execution.ThreadedTaskQueueExecutor$ExecutorRunnable$1.run(ThreadedTask QueueExecutor.java:116) jvm 1| at edu.emory.mathcs.backport.java.util.concurrent.Executors$RunnableAdapter.call(Executors.java:442) jvm 1| at edu.emory.mathcs.backport.java.util.concurrent.FutureTask.run(FutureTask.java:176) jvm 1| at edu.emory.mathcs.backport.java.util.concurrent.ThreadPoolExecutor$Worker.runTask(ThreadPoolExecutor.j ava:665) jvm 1| at edu.emory.mathcs.backport.java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java: 690) jvm 1| at java.lang.Thread.run(Thread.java:801) ArneD wrote: Hi all, first of all, congratulations to all Archiva developers for releasing the 1.0 version! I started playing around with it a little bit, and ran into the following problem: After starting up a fresh instance with default configuration, I copied parts of my existing repository to Archiva's default internal repository. Then I used the Scan Repository Now button on the repository administration page. Afterwards, all artifacts were visible on the Browse page. So far so good. I then copied some more groups of my existing repo to Archiva's default internal repository, and used Scan Repository Now once again. But this time, the Browse page did not show the newly added groups. This was quite surprising, as the log output of RepositoryScanner clearly showed that Archiva DID walk over these new artifacts, without errors or warnings. I tried to use the Update Database Now button, but still no effect. I finally touched one of the POM files in the missing groups, i.e. I gave the POM file a new timestamp. After doing Scan Repository Now again, the artifact appeared on the Browse page. When browsing into the artifact, however, I get the error message Unable to find project model for [...]. Seems like a bug to me? I know that my playground scenario may be a little bit unusual, but I would expect that Archiva should always be able to synchronize itself to the file system contents, regardless of the files' timestamps. Thanks - Arne -- View this message in context: http://www.nabble.com/Repository-scanning-problem-in-1.0--tf4897121.html#a14025508 Sent from the archiva-users mailing list archive at Nabble.com.
Re: Repository scanning problem in 1.0?
Wendy Smoak-3 wrote: On Nov 29, 2007 6:07 AM, ArneD [EMAIL PROTECTED] wrote: After starting up a fresh instance with default configuration, I copied parts of my existing repository to Archiva's default internal repository. Then I used the Scan Repository Now button on the repository administration page. Afterwards, all artifacts were visible on the Browse page. So far so good. I then copied some more groups of my existing repo to Archiva's default internal repository, and used Scan Repository Now once again. But this time, the Browse page did not show the newly added groups. This was quite surprising, as the log output of RepositoryScanner clearly showed that Archiva DID walk over these new artifacts, without errors or warnings. I tried to use the Update Database Now button, but still no effect. I finally touched one of the POM files in the missing groups, i.e. I gave the POM file a new timestamp. After doing Scan Repository Now again, the artifact appeared on the Browse page. When browsing into the artifact, however, I get the error message Unable to find project model for [...]. I've seen Unable to find project model for [...] in recent versions of Archiva, also using with existing repository contents, but have not been able to isolate it with publicly available repo data. Can you come up with steps to reproduce this, and attach zipped repository contents to a JIRA issue? It shouldn't take more than a few artifacts to provoke the problem, once you know the cause. http://jira.codehaus.org/browse/MRM Thanks, -- Wendy I filed http://jira.codehaus.org/browse/MRM-608 for the unable to find project model problem. Do you think the repository scanning problem as such, as described in my original mail, should be filed as a second issue? Thanks - Arne -- View this message in context: http://www.nabble.com/Repository-scanning-problem-in-1.0--tf4897121.html#a14026445 Sent from the archiva-users mailing list archive at Nabble.com.
Repository scanning problem in 1.0?
Hi all, first of all, congratulations to all Archiva developers for releasing the 1.0 version! I started playing around with it a little bit, and ran into the following problem: After starting up a fresh instance with default configuration, I copied parts of my existing repository to Archiva's default internal repository. Then I used the Scan Repository Now button on the repository administration page. Afterwards, all artifacts were visible on the Browse page. So far so good. I then copied some more groups of my existing repo to Archiva's default internal repository, and used Scan Repository Now once again. But this time, the Browse page did not show the newly added groups. This was quite surprising, as the log output of RepositoryScanner clearly showed that Archiva DID walk over these new artifacts, without errors or warnings. I tried to use the Update Database Now button, but still no effect. I finally touched one of the POM files in the missing groups, i.e. I gave the POM file a new timestamp. After doing Scan Repository Now again, the artifact appeared on the Browse page. When browsing into the artifact, however, I get the error message Unable to find project model for [...]. Seems like a bug to me? I know that my playground scenario may be a little bit unusual, but I would expect that Archiva should always be able to synchronize itself to the file system contents, regardless of the files' timestamps. Thanks - Arne -- View this message in context: http://www.nabble.com/Repository-scanning-problem-in-1.0--tf4897121.html#a14025501 Sent from the archiva-users mailing list archive at Nabble.com.
Re: Building project root as recursive build
Emmanuel Venisse wrote: ArneD a écrit : I noticed two problems with building the root of a projcet as a recursive build. I'm still on 1.1-beta-3, but at least from the release notes I could not see anything that this has been changed for 1.1-beta-4 or the 1.1 RC. First of all, when adding a Maven 2.0+ project using a POM url, with the checkbox For multi modules project, load only root as recursive build enabled, still all sub-modules get added to Continuum and must be removed manually afterwards. I can't remember if it was a problem inbeta-3 but I tested with beta-4 and 1.1 final (that should be released in few days) and I have only the parent pom added. Thanks, I'll try again with 1.1 after the release, and would file a bug in case it occurs again. The second problem is regarding automatic builds in case one of the dependencies has changed. Only the dependencies of the root module seem to be considered. The project information page on multi-module projects also only shows the dependencies of the root module. If you use the recursive mode without sub-modules, continuum doesn't know sub-modules so it doesn't know sub-dependencies. In future versions, you'll can add some dependencies between continuum projects to create links between them but not in 1.1 Wouldn't it be possible and desireable for Continuum to evaluate the modules section of the root POM and then the POMs of all relevant sub-modules in order to determine the correct dependencies? Thanks - Arne -- View this message in context: http://www.nabble.com/Building-project-root-as-recursive-build-tf4841777.html#a13855880 Sent from the Continuum - Users mailing list archive at Nabble.com.
Building project root as recursive build
I noticed two problems with building the root of a projcet as a recursive build. I'm still on 1.1-beta-3, but at least from the release notes I could not see anything that this has been changed for 1.1-beta-4 or the 1.1 RC. First of all, when adding a Maven 2.0+ project using a POM url, with the checkbox For multi modules project, load only root as recursive build enabled, still all sub-modules get added to Continuum and must be removed manually afterwards. The second problem is regarding automatic builds in case one of the dependencies has changed. Only the dependencies of the root module seem to be considered. The project information page on multi-module projects also only shows the dependencies of the root module. Am I missing something? Should I file an issue? Thanks - Arne -- View this message in context: http://www.nabble.com/Building-project-root-as-recursive-build-tf4841777.html#a13852345 Sent from the Continuum - Users mailing list archive at Nabble.com.
Re: Rights management for Proxy Connectors
Brett Porter wrote: I like this idea - can you file a feature request? I filed http://jira.codehaus.org/browse/MRM-579 Cheers - Arne -- View this message in context: http://www.nabble.com/Rights-management-for-Proxy-Connectors-tf4752902.html#a13602241 Sent from the archiva-users mailing list archive at Nabble.com.
Rights management for Proxy Connectors
Hello all, is it possible to configure Archiva in a way, that access to repo1 through a proxy connector is available only for a few super users (= repository managers)? In my scenario, normal users should only be able to access the artifacts available in the internal repository. In case they need a new artifact from repo1, they ask the repository manager to add it. The repository manager checks whether the artifact complies to e.g. company licensing rules, whether the metadata on repo1 is ok (which is mostly but not always the case), and so on. If everything is ok, he downloads the artifact including all dependencies. Afterwards the artifact should be available to everybody inside the company. The repository observer role seems to allow access to the proxied repositories as well, right? Thanks - Arne -- View this message in context: http://www.nabble.com/Rights-management-for-Proxy-Connectors-tf4752902.html#a13590815 Sent from the archiva-users mailing list archive at Nabble.com.
Continuum 1.1 release?
Hi Emmanuel and others, now that Maven 2.0.5 has been released, any chances that we see a Continuum 1.1 release soon? That would be great. Thanks - Arne -- View this message in context: http://www.nabble.com/Continuum-1.1-release--tf3238534.html#a9001442 Sent from the Continuum - Users mailing list archive at Nabble.com.
Download artifacts from one repo and upload to another
I'm looking for an easy way to download an artifact from one repository and deploy it to another, including all transitive dependencies. Ideally in just one step. Mirroring the whole repository is not a solution. How would you do it? Any hints what to look at? Thanks - Arne -- View this message in context: http://www.nabble.com/Download-artifacts-from-one-repo-and-upload-to-another-tf3181645s177.html#a8829481 Sent from the Maven - Users mailing list archive at Nabble.com. - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
Re: Lib dependencies in .war artifact
Morgovsky, Alexander (US - Glen Mills) wrote: Hi. I need some .jar's for compilation of a .war artifact, but I want to only include a subset of these in the WEB-INF/lib. If I set the subset of the ones for inclusion with the scope of runtime for each, they will not be used for compilation, and the compilation will fail. Thus, the question is, how do I include a subset of .jar's which both are for compilation and runtime, but exclude all the ones which are for compilation only? Thanks for your help. Try scope provided for those that you need for compilation only and scope compile for thos that you need for compilation and runtime. Regards, Arne -- View this message in context: http://www.nabble.com/Lib-dependencies-in-.war-artifact-tf2408437.html#a6717565 Sent from the Maven - Users mailing list archive at Nabble.com. - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
Re: Occassional NPEs while checking out through scm-local in scheduled b
Hello Emmanuel, I've found another NPE bug in DefaultContinuumScm which is easy to fix: http://jira.codehaus.org/browse/CONTINUUM-937 Regards, Arne Emmanuel Venisse wrote: I've seen that you closed issue CONTINUUM-780 in the meanwhile. Are there chances to replace just one module in my Continuum 1.0.3 installation by a snapshot version to get the fix? Thanks for a short hint in which module you fixed it. I don't think you can change the jar in 1.0.3 with a 1.1-SNAPSHOT. You can checkout the 1.0.3 tag, apply my patch on continuum-core and build it. -- View this message in context: http://www.nabble.com/Occassional-NPEs-while-checking-out-through-scm-local-in-scheduled-build-tf2289437.html#a6447722 Sent from the Continuum - Users mailing list archive at Nabble.com.
Re: Occassional NPEs while checking out through scm-local in scheduled b
Hello Emmanuel, I've seen that you closed issue CONTINUUM-780 in the meanwhile. Are there chances to replace just one module in my Continuum 1.0.3 installation by a snapshot version to get the fix? Thanks for a short hint in which module you fixed it. BTW, did you have a look at SCM-231? Regards, Arne ArneD wrote: Hello Emmanuel, i added a comment to the existing issue at http://jira.codehaus.org/browse/CONTINUUM-780 because it's the same stacktrace. Regards, Arne Emmanuel Venisse wrote: Can you file an issue? Emmanuel ArneD a écrit : It's Continuum 1.0.3, but I've updated maven-scm-api and maven-scm-provider-local (snapshot version maven-scm-provider-local-1.0-20060616.154545-4.jar). Arne Emmanuel Venisse wrote: What is your continuum version and your scm? Emmanuel ArneD a écrit : Hi, occasionally some of our builds are failing during scheduled builds. Forced builds are working fine. I see the following stacktrace: org.apache.maven.continuum.scm.ContinuumScmException: Cannot checkout sources. at org.apache.maven.continuum.scm.DefaultContinuumScm.updateProject(DefaultContinuumScm.java:276) at org.apache.maven.continuum.core.action.UpdateWorkingDirectoryFromScmContinuumAction.execute(UpdateWorkingDirectoryFromScmContinuumAction.java:58) at org.apache.maven.continuum.buildcontroller.DefaultBuildController.build(DefaultBuildController.java:166) at org.apache.maven.continuum.buildcontroller.BuildProjectTaskExecutor.executeTask(BuildProjectTaskExecutor.java:47) at org.codehaus.plexus.taskqueue.execution.ThreadedTaskQueueExecutor$ExecutorRunnable.run(ThreadedTaskQueueExecutor.java:103) at java.lang.Thread.run(Thread.java:534) Caused by: java.lang.NullPointerException at org.apache.maven.continuum.scm.DefaultContinuumScm.convertScmResult(DefaultContinuumScm.java:414) at org.apache.maven.continuum.scm.DefaultContinuumScm.updateProject(DefaultContinuumScm.java:244) ... 5 more It seems to be the following issue. A difference is that not all builds are failing, and the problems do not occur every time. http://jira.codehaus.org/browse/CONTINUUM-780 Any ideas how to resolve it yet? Is somebody working on it? Regards, Arne -- View this message in context: http://www.nabble.com/Occassional-NPEs-while-checking-out-through-scm-local-in-scheduled-build-tf2289437.html#a6412758 Sent from the Continuum - Users mailing list archive at Nabble.com.
Re: [m2] Include project version into generated site
Alan Mosely wrote: Hi, On a related note I would like to have the version included in the url for further transparency. Any ideas? We use the following distributionManagement entry in the project's pom.xml: distributionManagement site idmy-site/id urlfile://\\myserver\projectsites\${project.artifactId}-${project.version}/url /site /distributionManagement Regards, Arne -- View this message in context: http://www.nabble.com/-m2--Include-project-version-into-generated-site-tf2291271.html#a6386759 Sent from the Maven - Users mailing list archive at Nabble.com. - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
Re: [M2] Problem with surefire report
Luis Perea wrote: The surefire's report show me the same details for each diferent Test in my proyect, I've allready looked out the *.txt and *.xml files generated by surefire and the *.txt seems correct (shows the correct log for the TestCase executed) but the *.xml files has the same info for all the Tests. That's probably the same issue as described here: http://jira.codehaus.org/browse/MSUREFIRE-114 Regards, Arne -- View this message in context: http://www.nabble.com/-M2--Problem-with-surefire-report-tf2295962.html#a6386812 Sent from the Maven - Users mailing list archive at Nabble.com. - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
Re: Occassional NPEs while checking out through scm-local in scheduled build
Hello Emmanuel, i added a comment to the existing issue at http://jira.codehaus.org/browse/CONTINUUM-780 because it's the same stacktrace. Regards, Arne Emmanuel Venisse wrote: Can you file an issue? Emmanuel ArneD a écrit : It's Continuum 1.0.3, but I've updated maven-scm-api and maven-scm-provider-local (snapshot version maven-scm-provider-local-1.0-20060616.154545-4.jar). Arne Emmanuel Venisse wrote: What is your continuum version and your scm? Emmanuel ArneD a écrit : Hi, occasionally some of our builds are failing during scheduled builds. Forced builds are working fine. I see the following stacktrace: org.apache.maven.continuum.scm.ContinuumScmException: Cannot checkout sources. at org.apache.maven.continuum.scm.DefaultContinuumScm.updateProject(DefaultContinuumScm.java:276) at org.apache.maven.continuum.core.action.UpdateWorkingDirectoryFromScmContinuumAction.execute(UpdateWorkingDirectoryFromScmContinuumAction.java:58) at org.apache.maven.continuum.buildcontroller.DefaultBuildController.build(DefaultBuildController.java:166) at org.apache.maven.continuum.buildcontroller.BuildProjectTaskExecutor.executeTask(BuildProjectTaskExecutor.java:47) at org.codehaus.plexus.taskqueue.execution.ThreadedTaskQueueExecutor$ExecutorRunnable.run(ThreadedTaskQueueExecutor.java:103) at java.lang.Thread.run(Thread.java:534) Caused by: java.lang.NullPointerException at org.apache.maven.continuum.scm.DefaultContinuumScm.convertScmResult(DefaultContinuumScm.java:414) at org.apache.maven.continuum.scm.DefaultContinuumScm.updateProject(DefaultContinuumScm.java:244) ... 5 more It seems to be the following issue. A difference is that not all builds are failing, and the problems do not occur every time. http://jira.codehaus.org/browse/CONTINUUM-780 Any ideas how to resolve it yet? Is somebody working on it? Regards, Arne -- View this message in context: http://www.nabble.com/Occassional-NPEs-while-checking-out-through-scm-local-in-scheduled-build-tf2289437.html#a6388904 Sent from the Continuum - Users mailing list archive at Nabble.com.
Occassional NPEs while checking out through scm-local in scheduled build
Hi, occasionally some of our builds are failing during scheduled builds. Forced builds are working fine. I see the following stacktrace: org.apache.maven.continuum.scm.ContinuumScmException: Cannot checkout sources. at org.apache.maven.continuum.scm.DefaultContinuumScm.updateProject(DefaultContinuumScm.java:276) at org.apache.maven.continuum.core.action.UpdateWorkingDirectoryFromScmContinuumAction.execute(UpdateWorkingDirectoryFromScmContinuumAction.java:58) at org.apache.maven.continuum.buildcontroller.DefaultBuildController.build(DefaultBuildController.java:166) at org.apache.maven.continuum.buildcontroller.BuildProjectTaskExecutor.executeTask(BuildProjectTaskExecutor.java:47) at org.codehaus.plexus.taskqueue.execution.ThreadedTaskQueueExecutor$ExecutorRunnable.run(ThreadedTaskQueueExecutor.java:103) at java.lang.Thread.run(Thread.java:534) Caused by: java.lang.NullPointerException at org.apache.maven.continuum.scm.DefaultContinuumScm.convertScmResult(DefaultContinuumScm.java:414) at org.apache.maven.continuum.scm.DefaultContinuumScm.updateProject(DefaultContinuumScm.java:244) ... 5 more It seems to be the following issue. A difference is that not all builds are failing, and the problems do not occur every time. http://jira.codehaus.org/browse/CONTINUUM-780 Any ideas how to resolve it yet? Is somebody working on it? Regards, Arne -- View this message in context: http://www.nabble.com/Occassional-NPEs-while-checking-out-through-scm-local-in-scheduled-build-tf2289437.html#a6358643 Sent from the Continuum - Users forum at Nabble.com.
Re: Occassional NPEs while checking out through scm-local in scheduled build
It's Continuum 1.0.3, but I've updated maven-scm-api and maven-scm-provider-local (snapshot version maven-scm-provider-local-1.0-20060616.154545-4.jar). Arne Emmanuel Venisse wrote: What is your continuum version and your scm? Emmanuel ArneD a écrit : Hi, occasionally some of our builds are failing during scheduled builds. Forced builds are working fine. I see the following stacktrace: org.apache.maven.continuum.scm.ContinuumScmException: Cannot checkout sources. at org.apache.maven.continuum.scm.DefaultContinuumScm.updateProject(DefaultContinuumScm.java:276) at org.apache.maven.continuum.core.action.UpdateWorkingDirectoryFromScmContinuumAction.execute(UpdateWorkingDirectoryFromScmContinuumAction.java:58) at org.apache.maven.continuum.buildcontroller.DefaultBuildController.build(DefaultBuildController.java:166) at org.apache.maven.continuum.buildcontroller.BuildProjectTaskExecutor.executeTask(BuildProjectTaskExecutor.java:47) at org.codehaus.plexus.taskqueue.execution.ThreadedTaskQueueExecutor$ExecutorRunnable.run(ThreadedTaskQueueExecutor.java:103) at java.lang.Thread.run(Thread.java:534) Caused by: java.lang.NullPointerException at org.apache.maven.continuum.scm.DefaultContinuumScm.convertScmResult(DefaultContinuumScm.java:414) at org.apache.maven.continuum.scm.DefaultContinuumScm.updateProject(DefaultContinuumScm.java:244) ... 5 more It seems to be the following issue. A difference is that not all builds are failing, and the problems do not occur every time. http://jira.codehaus.org/browse/CONTINUUM-780 Any ideas how to resolve it yet? Is somebody working on it? Regards, Arne -- View this message in context: http://www.nabble.com/Occassional-NPEs-while-checking-out-through-scm-local-in-scheduled-build-tf2289437.html#a6362234 Sent from the Continuum - Users mailing list archive at Nabble.com.
[m2] Include project version into generated site
Hi, I would like to have the project version included into the generated site, e.g. into the title, to make this information transparent to the user directly on the welcome page. I tried to use ${project.version} in the site.xml descriptor, but this is not replaced. Only ${project.name} seems to be working. Any hints? Thanks, Arne -- View this message in context: http://www.nabble.com/-m2--Include-project-version-into-generated-site-tf2291271.html#a6363605 Sent from the Maven - Users mailing list archive at Nabble.com. - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
Re: Update over scm-local does not delete files removed from source dir
Emmanuel Venisse wrote: Do you think the suggested enhancement for scm-local would be useful and make sense? Not only in combination with ClearCase dynamic views but for other usage as well, e.g. testing. yes. Do you want to implement it? Yes, I can try to do so and then provide a patch. Regards, Arne -- View this message in context: http://www.nabble.com/Update-over-scm-local-does-not-delete-files-removed-from-source-dir-tf2257460.html#a6303945 Sent from the Continuum - Users forum at Nabble.com.
Re: Update over scm-local does not delete files removed from source dir
Thanks for your answer, David. continuum-3 wrote: my Continuum installation is accessing a file system directory using scm-local provider to get sources (BTW, the directory is on a ClearCase dynamic view). Updates are working fine as long as files are only changed or added. But when files are removed from the source directory, they still exist in the target directory. Especially after refactoring activities, this leads to build errors. Make sure you are runnning clean goal in maven. When you do a clean it should delete this info from the target directories. And that will fix your problem. This is what I am doing. The problem is: My project source dir resides on a dynamic view on drive Z:\VOB\some\where. My Continuum working directory is e.g. D:\continuum-work\99. What Continuum does (with help of scm-local) is to copy all files from Z:\VOB\some\where to D:\continuum-work\99 before running the build. If someone removes an outdated class from ClearCase, it won't be on Z:\VOB\some\where any longer. As scm-local currently does not delete anything from the checkout directory, the outdated class will still be there at d:\continuum-work\99. The clean goal will only delete d:\continuum-work\99\target. BTW, you'll speed up your build if you get maven to put the target directories somewhere else, outside of the dynamic view, as dynamic views can be slow. Did you manage to let Continuum operate directly on a dynamic view, in my example on Z:\VOB\some\where, without copying to a working directory? Probably you are only talking about Maven stand-alone usage. Then, I agree, it is no problem, as long as you are running the clean goal. Regards, Arne -- View this message in context: http://www.nabble.com/Update-over-scm-local-does-not-delete-files-removed-from-source-dir-tf2257460.html#a6309477 Sent from the Continuum - Users forum at Nabble.com.
Re: Update over scm-local does not delete files removed from source dir
Emmanuel Venisse wrote: thanks a lot for your answer. A ClearCase dynamic view looks like a normal filesystem, so I think scm-local in principle is a fine solution. I think there wouldn't be much what a dedicated dynamic-view support in the Clearcase SCM provider could add. Maybe it looks like a normal filesystem, but I think the clearcase server know if files are deleted and update your local copy, right? ClearCase of course has the information. But the problem is that a ClearCase dynamic view resides on a virtual network share, e.g. \\view\some_view. You can map it to a drive letter under windows, e.g. Z:, but I don't see a way to define it in Continuum's working directory. So the files have to be copied from the dynamic view to the working directory - and that's what scm-local is doing. Couldn't the scm-local adapter consider all files that are in the checkout dir but not in the source dir as deleted? This should be easy to implement and do the job. We can't, because some users (or the build) add some files in the checkout directory like the target directory and they don't want to remove them at each build. That's true. Maybe we could enhance scm-local to keep its own metadata? In particular, scm-local could maintain a simple file, say .maven-scm-local, that contains as plain text the list of files in the source directory, as seen during the last checkout or update operation: - During checkout, the file .maven-scm-local is created in the checkout base directory. Its just a plain text file containing the list of files that have been checked out. - The update command looks for the file. If it is there, it compares the contents of that file to the current source directory contents (including subdirs). All files that are in .maven-scm-local but are no longer in the source dir, have been deleted in the source dir. The update command therefore removes them from the checkout dir. - If for whatever reason .maven-scm-local is not there, the update command won't delete any files. That way, we're backwards compatible. - After completing the update process, the update command rewrites the .maven-scm-local metadata file. - Even the changelog command can interpret .maven-scm-local - For add and checkin commands, I don't think that changes are needed. What do you think? Regards, Arne -- View this message in context: http://www.nabble.com/Update-over-scm-local-does-not-delete-files-removed-from-source-dir-tf2257460.html#a6281593 Sent from the Continuum - Users forum at Nabble.com.
Re: Update over scm-local does not delete files removed from source dir
Emmanuel Venisse wrote: Are you sure it isn't possible to checkout sources in a specific folder with cleartool? It is possible with snapshot views. My ClearCase know-how is limited, but I am quite sure it is not possible with dynamic views. Do you think the suggested enhancement for scm-local would be useful and make sense? Not only in combination with ClearCase dynamic views but for other usage as well, e.g. testing. Regards, Arne That's true. Maybe we could enhance scm-local to keep its own metadata? In particular, scm-local could maintain a simple file, say .maven-scm-local, that contains as plain text the list of files in the source directory, as seen during the last checkout or update operation: - During checkout, the file .maven-scm-local is created in the checkout base directory. Its just a plain text file containing the list of files that have been checked out. - The update command looks for the file. If it is there, it compares the contents of that file to the current source directory contents (including subdirs). All files that are in .maven-scm-local but are no longer in the source dir, have been deleted in the source dir. The update command therefore removes them from the checkout dir. - If for whatever reason .maven-scm-local is not there, the update command won't delete any files. That way, we're backwards compatible. - After completing the update process, the update command rewrites the .maven-scm-local metadata file. - Even the changelog command can interpret .maven-scm-local - For add and checkin commands, I don't think that changes are needed. What do you think? Regards, Arne -- View this message in context: http://www.nabble.com/Update-over-scm-local-does-not-delete-files-removed-from-source-dir-tf2257460.html#a6289204 Sent from the Continuum - Users forum at Nabble.com.
Update over scm-local does not delete files removed from source dir
Hello everybody, my Continuum installation is accessing a file system directory using scm-local provider to get sources (BTW, the directory is on a ClearCase dynamic view). Updates are working fine as long as files are only changed or added. But when files are removed from the source directory, they still exist in the target directory. Especially after refactoring activities, this leads to build errors. Seems like a bug to me... Or am I missing something? Many thanks, Arne -- View this message in context: http://www.nabble.com/Update-over-scm-local-does-not-delete-files-removed-from-source-dir-tf2257460.html#a6261871 Sent from the Continuum - Users forum at Nabble.com.
Re: Update over scm-local does not delete files removed from source dir
Emmanuel Venisse wrote: scm-local was developped for tests and it's used in some case when a scm provider isn't available like dynamic-view support for Clearcase. It's a very simple provider that copy file from sources directory, but it can't know if some files are deleted because it doesn't have some metadata that can inform it. The only possibility you have for the moment is to remove the checkout directory and run a new build. But the best way would be to add dynamic-view support in clearcase provider. Only clearcase users will can add this features because we don't know clearcase and we don't have access to a clearcase server. Hello Emmanuel, thanks a lot for your answer. A ClearCase dynamic view looks like a normal filesystem, so I think scm-local in principle is a fine solution. I think there wouldn't be much what a dedicated dynamic-view support in the Clearcase SCM provider could add. Couldn't the scm-local adapter consider all files that are in the checkout dir but not in the source dir as deleted? This should be easy to implement and do the job. Best regards, Arne -- View this message in context: http://www.nabble.com/Update-over-scm-local-does-not-delete-files-removed-from-source-dir-tf2257460.html#a6267099 Sent from the Continuum - Users forum at Nabble.com.
Re: [POLL] Why switch to Maven?
Eric Redmond wrote: Any more reasons? Care to expand these ideas? Hi Eric, for corporate users, I believe there are some additional issues to the ones already named. Especially in large companies, it is often unacceptable to let users download artefacts directly from an Internet repository, even not through a proxy. Companies need to have full control over all artifacts used in their build processes, that means that only artefacts that have been explicitly released by someone should be used. So when you start using Maven, you want to immediately set up an internal repository which, I believe, is still a too complicated task. This is especially true for setting up an internal plugin repository. We did not find a comfortable way to initially fill an internal plugin repository. You should be able to say fill my plugin repository with all required JARs for the plugins x, y and z. Our workaround was to set up a Maven reference installation zip, that already has all required plugin JARs in its local repository, so that users who unzip that archive have everything they need. Another point is, plugin JARs and production JARs should be separated. This is already possible for remote repositories, but it should be possible for the local repository as well. Tool JARs like xerces etc. are used both by plugins and by projects. But just because some plugin needs a JAR file, corporate users do not want to make that JAR available for builds. Corporate users just don't want to get tools mixed up with production software. Anyway, Maven is a great piece of software, full of great ideas. But some things still need to be improved before Maven has chances to achieve a wide acceptance. Arne -- View this message in context: http://www.nabble.com/-POLL--Why-switch-to-Maven--tf2185174.html#a6078097 Sent from the Maven - Users forum at Nabble.com. - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
Re: [POLL] Why switch to Maven?
Eric Redmond wrote: Especially in large companies, it is often unacceptable to let users download artefacts directly from an Internet repository, even not through a proxy. Companies need to have full control over all artifacts used in theirE build processes, that means that only artefacts that have been explicitly released by someone should be used. I don't understand why a Proxy is unacceptible. The administrator of the Proxy can - at any time- disconnect the Proxy from the internet. In the company that I helped setting up a Maven-based build environment, a public site like ibiblio.org is considered a potentially unsafe source. Like it or not. Only JARs that have been approved internally may be used for production. (BTW, this was within finance industry which is partly quite sensitive at such points.) Practically, that meant we had to set up an internal remote repository and deploy all existing JARs to it. This worked fine in the end, but felt difficult in the beginning - probably due to the lack of good documentation and examples. The bigger problem was setting up an internal **plugin** repository: This is especially true for setting up an internal plugin repository. We did not find a comfortable way to initially fill an internal plugin repository. You should be able to say fill my plugin repository with all required JARs for the plugins x, y and z. Our workaround was to set up a Maven reference installation zip, that already has all required plugin JARs in its local repository, so that users who unzip that archive have everything they need. Why can't you solve this with an in-house remote repository, rather than zipping up jars for local repos? You can do it with an in-house remote plugin repository, but how would you populate it with all plugins that you need? Ok, you can set up a proxy for this purpose. Then, you somehow have to make sure that Maven downloads all relevant plugins (this is not trivial either). Afterwards you can disconnect the proxy from Internet, or use the proxy's cache as your internal plugin repository. (Keeping the proxy alive and connected to the Internet might be unacceptable because you want to evaluate new plugins before you release them for internal usage.) When I wanted to use Maven-proxy for this purpose, I soon encountered problems because it did not support NTLM authentication, so I could not get through the firewall. So we helped us with zipping up a local repo as a workaround. Anyway, I think that the whole task of setting up an internal plugin repository is too complicated. We don't want to use a proxy, so why do we have to set it up. When talking about tools, companies are used to download the tool from the vendor's site or install it from CD and then just use it. This is not possible with Maven, you need an Internet connection during run-time, at least until you have managed to set up an internal plugin repository. Another point is, plugin JARs and production JARs should be separated. This is already possible for remote repositories, but it should be possible for the local repository as well. Tool JARs like xerces etc. are used both by plugins and by projects. But just because some plugin needs a JAR file, corporate users do not want to make that JAR available for builds. Corporate users just don't want to get tools mixed up with production software. Why does your build software get mixed up with your production software? Using the release or dependency plugins will not bundle build tools with the jars to be released. Maybe it becomes clearer with an example: Let's say, your build uses plugin X which has a dependency to Xerces. So you will have Xerces in your local repository after running your build, because it is downloaded from the internal plugin repository to your local repository. Thereafter, you are able to declare a dependency to Xerces in your project, and the build will run through - even if Xerces has not been released to the internal remote repository, and there's no connection to the Internet. Builds should only be able to use artifacts that have explicitly been released to the internal remote repository. Nothing else. Even not JARs from an internal plugin repository. Anyway, Maven is a great piece of software, full of great ideas. But some things still need to be improved before Maven has chances to achieve a wide acceptance. Unless I'm really misunderstanding the problems, I haven't seen a problem you've had that is inherent to Maven (except possibly by creating a mirror repository with a set of dependencies). Did my points get a little bit clearer now? But don't get me wrong, I do not consider these points real showstoppers. We did set up a Maven-based build environment succesfully in the end, and it works quite well. Regards, Arne -- View this message in context: http://www.nabble.com/-POLL--Why-switch-to-Maven--tf2185174.html#a6084903 Sent from the Maven
Re: [POLL] Why switch to Maven?
cstamas wrote: Right. Take Proximity for example, since it is not JUST proxy. If you visit the demo site, you will see that proximity is able to PROXY repositories but also to HOST them only. Furthermore, you can just take offline (offline == do not touch remote peer!) or take unavailable (unavailable == refuse all requests) a repo from proximity by a switch. I agree, Proximity is promising, and I will definitely take a further look. I first heard about it after initially setting up the Maven build environment. Eric's question was, what is people preventing from using Maven. The points that I named did not prevent me from using Maven, neither from recommending its usage. I just want to point out things that corporate users are concerned of. All these things are solveable. The question is, how much do you get out-of-the-box and how easy is it to use. And for Xerces example: IF you have a developer who declares an obiously existing (it's in repo downloaded from plugn repo) artifact which is obviously forbidden (it is not in company repo), than you have a HR problem. No software will ever overcome human stupidity. Nor IT rigidness. The reality in many companies is that you have people with very different skills and experiences in IT departments. The idea of build automation is not only to make life simpler, but also to help people to avoid mistakes. Build automation should be easy-to-use for people even who use it the first time. So, in the Xerces example, one might just not be aware of the fact that Xerces is not in the company repository, but only in the plugin repository. Regards, Arne -- View this message in context: http://www.nabble.com/-POLL--Why-switch-to-Maven--tf2185174.html#a6088427 Sent from the Maven - Users forum at Nabble.com. - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
Re: [POLL] Why switch to Maven?
After all, if can't trust your team to stick to approved versions of artifacts how can you trust them to write your precious business code? I think it's not a question of mistrusting people, but a question of how can you help people to avoid mistakes. Even if they use a Maven-based build for the first time and don't have a M.Sc. in Computer Science or x years of experience. If the team *wants* to used unapproved versions of artifacts, of course, they will be able to do it. But you should help them to not do it by mistake. So, how to you verify instead of lock? You have a parent pom that declares and defines all versions of artifacts and plugins and in your module poms you declare that you want a plugin but you provide no version information. Then your configuration team only needs to check the parent pom against the internal standards. Nice suggestion, I'll think about it for the customer's environment. Anyway, if you talk about ease-of-use and quick adoption, this is not simple enough. I am sure that these issues prevent some people from using Maven - and that was Eric's question. NTLM support would help but hey, it's a closed proprietary undocumented authentication scheme from Microsoft, you can't expect everyone to support it. Again, Eric's question was what is people preventing from using it. The reality in many companies is, they do use NTLM. As a workaround you can use NTLMAPS on sourceforge as a local proxy for any apps that don't know how to support NTLM. Possible, but unnecessarily complicated, and far away from out-of-the-box. Regards, Arne -- View this message in context: http://www.nabble.com/-POLL--Why-switch-to-Maven--tf2185174.html#a6088629 Sent from the Maven - Users forum at Nabble.com. - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
maven-dependency-plugin - (Beta-)Release planned?
Are there plans for releasing maven-dependency-plugin? The last snapshot version is from April: http://people.apache.org/maven-snapshot-repository/org/apache/maven/plugins/maven-dependency-plugin/2.0-SNAPSHOT/ If 2.0 is not yet ready, a beta release would be great. Thank you, Arne -- View this message in context: http://www.nabble.com/maven-dependency-plugin---%28Beta-%29Release-planned--tf2182224.html#a6035134 Sent from the Maven - Users forum at Nabble.com. - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
maven-dependency-plugin - Why is scope parameter read-only?
I am trying to use maven-dependency-plugin (latest snapshot version) using the following configuration: plugin groupIdorg.apache.maven.plugins/groupId artifactIdmaven-dependency-plugin/artifactId version2.0-SNAPSHOT/version executions execution idcopy_jars_to_webinf_lib/id phaseinitialize/phase goals goalcopy-dependencies/goal /goals configuration outputDirectory${basedir}/WebContent/WEB-INF/lib/outputDirectory scopecompile/scope /configuration /execution /executions /plugin However, I get the error message: [INFO] Error configuring: org.apache.maven.plugins:maven-dependency-plugin. Reason: ERROR: Cannot override read-only parameter: scope in goal: dependency:copy-dependencies Why is scope read-only? (See AbstractDependencyFilterMojo.scope). Thanks, Arne -- View this message in context: http://www.nabble.com/maven-dependency-plugin---Why-is-%22scope%22-parameter-read-only--tf2182289.html#a6035345 Sent from the Maven - Users forum at Nabble.com. - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
[M2] maven-clean-plugin and multi-module build
I have a multi-module build. In one of the sub-modules I use the maven-clean-plugin in the pom.xml as follows: plugin artifactIdmaven-clean-plugin/artifactId configuration filesets fileset directoryWebContent/WEB-INF/lib/directory includes include*/include /includes /fileset /filesets /configuration /plugin The WebContent/WEB-INF/lib folder gets cleaned, when I call mvn clean on the sub-module. But it does not get cleaned when I call mvn clean on the main module. Should I file a JIRA issue? Or am I making a mistake? Is there any workaround available? Thanks, Arne -- View this message in context: http://www.nabble.com/-M2--maven-clean-plugin-and-multi-module-build-tf2177477.html#a6020540 Sent from the Maven - Users forum at Nabble.com. - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
RE: [m2] Updates of transitive dependencies not working?
Jörg Schaible wrote: ArneD wrote on Thursday, August 24, 2006 2:43 PM: I defined two repositories in the settings.xml, both with updatePolicy set to always. My project A has a dependency to version 5.0-SNAPSHOT of a JAR B. That JAR B has a dependency to version 1.6 of another JAR C. In my local repository there's an outdated version 1.6 of JAR C (i.e. version 1.6 has been redeployed after a bug has been found). The problem is: During my build of project A Maven is looking for an update of JAR B, but NOT of JAR C. Is this a bug or am I missing some setting? Yes. You cannot upgrade a final version! Therefore it is final. Maven will *never* look for an update. That's what snapshots are for I know that you normally should use snapshot versions for that. Anyway, Maven *does* allow to overwrite an existing version in the repository by re-deploying it. To make builds repeatable, I believe, Maven has to look for updates, even for released versions, not only snapshot versions. Isn't that what the repositoryreleaseupdatePolicyalways/... setting is for? (although it does also not work because of a bug). Do you know the bug number? Thank you! Arne -- View this message in context: http://www.nabble.com/-m2--Updates-of-transitive-dependencies-not-working--tf2158398.html#a5963829 Sent from the Maven - Users forum at Nabble.com. - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
RE: [m2] Updates of transitive dependencies not working?
Thank you, Jörg. I filed issue MNG-2528. Arne -- View this message in context: http://www.nabble.com/-m2--Updates-of-transitive-dependencies-not-working--tf2158398.html#a5964861 Sent from the Maven - Users forum at Nabble.com. - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
Maven 2 and JUnit-Tests for Eclipse Plugins
Has anybody set up Maven 2 builds for Eclipse (PDE) plugin projects and has managed to integrate JUnit in-container tests for Eclipse-Plugins into the Maven build lifecycle? I would be very much interested in some best practices. Thank you! Arne -- View this message in context: http://www.nabble.com/Maven-2-and-JUnit-Tests-for-Eclipse-Plugins-tf1942500.html#a5324188 Sent from the Maven - Users forum at Nabble.com. - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
Re: ClearCase SCM and Continuum
Wim Deblauwe wrote: I found another workaround by building the super-module recursively: Delete all sub-modules. Change the super module definition and remove the --non-recursive option from the build targets. Maybe we need an option in Continuum so the user can select if he wants his submodules as seperate modules (default now) or as just 1 project without the --non-recursive option? Yes, I believe that would be helpful. By the way, do you know has anybody planned to implement the checkout by tag functionality for ClearCase, so that the Maven release plugin can be used? you can perhaps help Wim and us if you can provide this feature in maven-scm. That would be great if you could do that. I started the ClearCase implementation because I hoped we would be moving to M2 soon, but it still has not happend. Currently, I can't spend any more time on the SCM implementation of ClearCase to do that, so if you could help out, it would be great! I am not sure if I can do something here. I am currently helping a customer in setting up a Maven-based build environment. (They use ClearCase at the moment but think about moving to SVN which would make it obsolete.) My own ClearCase knowledge is very limited, and I also do not have access to a ClearCase environment when I'm not on-site with the customer. But I'll see if I can do something. Can't promise though. And, at last, one suggestion. I believe the ClearCase integration would be even better, if the SCM url would contain all necessary information (like with CVS, SVN, ...). To define the config spec as an external file adds unnecessary complexity for the users. Anyway, thank you and all Maven / Continuum developers! You've done a great job. But for a broad acceptance of Maven, I believe, a simplification of some issues would be helpful. All proposals are welcome. Yes, how would you do that in 1 line? Most config specs are at least 2 lines, so that is why we need an external file. But if you can think of something, please do. My idea was that maybe the config spec could be generated by the SCM plugin. That would somehow anyway be necessary in order to make the checkout from a tag functionality work. But, as said, I currently only have limited ClearCase knowledge, and I am not sure what can be specified in the config spec and how much flexibility is needed there. Anyway, the trivial cases I've seen so far, were mostly the same two lines that are needed for checking out the main trunk: element * CHECKEDOUT element * /main/LATEST Maybe a config spec can generated for most cases (for checking out from the main trunk, from branches and from tags). The SCM url could look similiar to a SVN SCM url in that case. Optionally a reference to ones own config spec could be specified with a SCM url similiar to the current ClearCase SCM url. Regards, Arne -- View this message in context: http://www.nabble.com/ClearCase-SCM-and-Continuum-t1812397.html#a4981893 Sent from the Continuum - Users forum at Nabble.com.
Re: ClearCase SCM and Continuum
Thank you, Wim! Wim Deblauwe wrote: 1) This is the way that ClearCase works, there is nothing I know we can do about it. Maybe something could be changed in Continuum and the POM schema, so that the POM location relative to the SCM checkout directory could be specified within the POM. 2) I guess the workaround here would be to add the submodules individually. You can remove the view that was created via cleartool or from the clearcase home base. I found another workaround by building the super-module recursively: Delete all sub-modules. Change the super module definition and remove the --non-recursive option from the build targets. By the way, do you know has anybody planned to implement the checkout by tag functionality for ClearCase, so that the Maven release plugin can be used? And, at last, one suggestion. I believe the ClearCase integration would be even better, if the SCM url would contain all necessary information (like with CVS, SVN, ...). To define the config spec as an external file adds unnecessary complexity for the users. Anyway, thank you and all Maven / Continuum developers! You've done a great job. But for a broad acceptance of Maven, I believe, a simplification of some issues would be helpful. Regards, Arne -- View this message in context: http://www.nabble.com/ClearCase-SCM-and-Continuum-t1812397.html#a4956443 Sent from the Continuum - Users forum at Nabble.com.
Automatically deploy snapshot bulds to remote repo
Hi, is there a way of telling Continuum to make nightly builds only for snapshot versions? Particularly, I want to have the nightly snapshot builds automatically deployed to the snapshot repository. This is why I would change the build goals from clean install to clean deploy. Projects that have a non-snapshot version (i.e. projects that are currently not under development) should not be deployed. Thanks! Arne -- View this message in context: http://www.nabble.com/Automatically-deploy-snapshot-bulds-to-remote-repo-t1798808.html#a4902069 Sent from the Continuum - Users forum at Nabble.com.
Re: Automatically deploy snapshot bulds to remote repo
Thank you, Emmanuel. Emmanuel Venisse wrote: yes, goals to use is clean deploy, but continuum don't know what is a project in dev and a released project. Generally, when we release a project, the code is tagged and the code in trunk is updated to an incremented snapshot version. clean deploy works well if you always have a snapshot version under development. But that's not the case for every project. Sometimes you release a version, and do not have plans for a future release yet. You do not know, will the next rel. be a major or a minor release? Wiill there be a next release at all? etc. At the moment, I only see one solution, but I do not like it very much: - Only keep snapshot versions in Continuum. - Set build goals to clean deploy - Make sure that Continuum only deploys snapshot versions by defining only the snapshot repository in the distributionManagement section of the pom. The release repository would be defined within a profile section that has to explicitly be enabled, e.g. by parameter -Drelease=true. That way the deployment of the release version can only be done manually, when the user explicitly sets the -Drelease=true parameter. If, by mistake, a release version is in Continuum, the build will fail because of the missing distributionManagement information. Do you have a better idea? Thanks again Arne -- View this message in context: http://www.nabble.com/Automatically-deploy-snapshot-bulds-to-remote-repo-t1798808.html#a4903614 Sent from the Continuum - Users forum at Nabble.com.
Maven 2 and RAD 6
I am trying to integrate my Mavenized web projects with RAD6. Even though I was searching the list archives up and down, I still find this difficult, because the information on the list is very fragmented and partly outdated. I believe a kind of How-To documentation would be very helpful. I tried to follow the instructions at: http://www.nabble.com/Best+Practice+-+Maven+with+WSAD+or+RAD6-t966147.html#a2509735 All the JARs should be located in the EAR project's lib dir - not in the WAR project's WEB-INF/lib. Questions: 1. Even though I specified warSourceExcludesWEB-INF/lib/*/warSourceExcludes in the pom.xml as described in the thread above, all the JARs are still copied to WEB-INF/lib. 2. How can I achieve that during the build of the EAR project, all the JARs are automatically copied to the EAR's *source* directory, so that RAD can find the JARs thereafter. I am searching for something like mvn war:inplace, but for the EAR project. Any help would greatly be appreciated! Thanks, Arne -- View this message in context: http://www.nabble.com/Maven+2+and+RAD+6-t1711035.html#a4645535 Sent from the Maven - Users forum at Nabble.com. - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
Internal plugin repository - how to populate?
In many corporate environments it is unacceptable to let users download Maven plugins automatically and directly from the Internet (even not through a proxy). Therefore, it is necessary to set up an internal plugin repository. Question: Is there an easy way how to initially populate the internal repository with all the maven plugins needed? I considered the following ways: 1. mvn deploy:deploy-file for all the plugin jars. Problem: This seems to be a rather inconvenient approach with lots of manual work. 2. First popule a local repository by running all desired Maven goals, and then copy the local repository contents to the internal remote repository. Problem: As far as I understood, this wouldn't work, because a remote repository contains some extra XML definition files, that are not available in the local repository. Any other ideas? Thanks in advance, Arne -- View this message in context: http://www.nabble.com/Internal+plugin+repository+-+how+to+populate--t1697509.html#a4606575 Sent from the Maven - Users forum at Nabble.com. - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
[2.0.4] Bug? Profile not used when running outside a Maven project
I want to roll out Maven in a corporate environment with an internal plugin repository. My approach is to define and activate a profile in the global settings.xml, which normally works fine. Problem: When running Maven in a directory that does not contain a pom.xml, the profile from settings.xml does not become active: Maven does not use the pluginRegistry setting and connects to the public repository at repo1.maven.org instead. Seems like a bug to me? Thanks! Arne -- View this message in context: http://www.nabble.com/-2.0.4-+Bug-+Profile+not+used+when+running+outside+a+Maven+project-t1699843.html#a4612873 Sent from the Maven - Users forum at Nabble.com. - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]