I downloaded the source code for the EMF based UML2 Plugin and changed a class in the org.eclipse.uml2.uml.edit project to remove special characters when returning string representations. Now when I export the projects and place the jar files either in the dropins directory or replace my current uml2 plugin jar files in plugins directory, The UML files are no longer recognized, in short my modified plugin does not install correctly (no error is thrown and I can see the files being picked up under Plugins->Target Platform) .
However, When I run the plugin as an eclipse application (from the workspace) I can see the changes I made being reflected in the new instance of eclipse.
What can I do to ensure that the plugin installs correctly?
Is there a documented procedure of how to build the uml2 plugin (or any comparable plugin) after modification?
Select the project and open the context menu. There is an entry PDE near the bottom of the menu. In there, you can find an entry to build the plugin for deployment. This gives you the features and plugins directory with the fixed files. Copy both into your Eclipse install.
Unless the UML2 plugins require some kind of magic build script, exporting the one plugin you changed and overwriting the original in your Eclipse installation should be the easiest solution. One potential problem which comes to mind is conflicting plugin version numbers: make sure you don't have two identical versions of your modified plugin in your Eclipse installation.
When debugging plugins which apparently don't work properly at runtime, I always look at Help > About Eclipse Platform > Configuration Details. This lists all the plugins found by Equinox during startup, along with their status (see the Javadoc of the org.osgi.framework.Bundle interface for explanation).
I faced the exact same problem as you describe here . I dont have any answer to your problem but i am sharing what worked for me .
I created a local update site of the plugin on my system. Create update site for your plug-in article explains very very nicely the steps needed to accomplish this .
Related
I'm in the early stages of creating a new Eclipse plug-in in Eclipse 2019-06. I created the plug-in using the "Hello World" wizard option and created a debug configuration that successfully launches a secondary workbench for testing. I can click on my new menu option and step into my new code.
I want to use a modified version of org.eclipse.jdt.junit.wizards.NewTestCaseWizardPageOne, so I added the package to my project and edited NewTestCaseWizardPageOne. However, when I try to debug, the debugger asks for a source location. It seems to want to load the file from the local repository (.p2/pool/plugins/org.eclipse.jdt.junit_3.11.400.v20190510-0840.jar), rather than my edited version. Why is that?
There must be some kind of classpath problem, but don't versions of files in the project have precedence over those in the dependent plugins? I've looked at a number of similar questions (1, 2) and other sources, but I haven't yet found the answer. Please help.
Following Alexander Federov's suggestion, I converted my plugin into a fragment. This was fairly easy to do following the advice from this StackOverflow page. The main changes were:
Renamed plugin.xml to fragment.xml and changed the top level xml
element from plugin to fragment.
Added a Fragment-Host entry to MANIFEST.MF
The key advantage that this provided is discussed in an Eclipse Wiki
page:
... a fragment appears much the
same as a normal plug-in. A fragment can specify libraries,
extensions, and other files. When it is loaded by the platform loader,
a fragment is logically, but not physically, merged into the host
plug-in. The end result is exactly the same as if the fragment's
manifest were copied into the plug-in manifest, and all the files in
the fragment directory appear as if they were located in the plug-in's
install directory. Thus, a runtime library supplied by a fragment
appears on the classpath of its host plug-in. In fact, a Java class in
a fragment can be in the same package as a class in the host and will
even have access to package-visible methods on the host's classes.
The last part having to do with access to package-visible methods was what I needed. Thanks, Alexander!
I'm trying out Java 9 Jigsaw module system (no module experience yet) and would like to use it for capsuling the classes within my project, but it's confusing.
According to this article it should be possible to have multiple modules within ONE project. I made a new project in Eclipse Oxygen (Java 9 is supported) with the same structure as shown in the article. But Eclipse keeps telling me that I must not have more than one module-info.java in a project.
I really don't know how to tell Eclipse that it should use the "multi-module-mode". And I really would appreciate not having to create a new project for every single module.
This works:
This not:
But according to this article something like that should work:
And how about deployment of a modularized project with Eclipse? There is nothing to see about the new jmod extension. Do I still export it as a runnable JAR file like before?
Notice that my questions refer to working with the IDE (no command line, I mean with an IDE that should be possible, right?) Thank you for enlightening me.
Currently, Eclipse requires you to create a separate project for each module (e. g. because each module has its own Java Build Path).
To understand this design decision, consider that Java modules correspond to OSGi bundles / Eclipse plug-ins and it has always been to have a separate project for each bundle/plug-in. If you come from the Maven world, you would probably expect a deeper folder structure instead. But modules are self-contained and combining several modules into one project would only add an additional folder level without meaning. However, Eclipse supports nested projects and so-called working sets if you need an additional folder level.
Exporting modules as images is planned for Eclipse 2019-03 (4.11), on March 20, 2019 (see Eclipse bug 518445). Exporting modules as JARs that can be used on the modulepath (-m) already works (see my video).
I don't know if this question is still open for an answer, but you can solve this problem by simply removing all source folders on the build path. At least this works for Eclipse 2021-12 version.
As you can see this is a demo project from the Official Gradle Guide Book and it has multiple modules. Each module has its own module-info.java.
project structure in IntelliJ IDEA
If I open this project in Eclipse it will give me the 'duplicated entries on module-info.java' error.
Eclipse shows the error
But if I delete all the source folders on the build path, the error is gone and the project can be built and run without problem.
project properties: Java build path
The only problem is that you have to build the project with Gradle so that it will produce the .jar of each module and you have to include them in the libraries later.
include all the .jar in libraries
I think this is probably the same solution mentioned by howlger above.
I need to configure my eclipse product to load all features/plugins in features/plugins directory. Now my product loads just plugins that are listed in metadata, so if I export some plugin directly from eclipse tool as "deployable plugins" it's loaded in my product, but if I put a plugin jar directly in plugins directory It doesn't appear (the same if I work with features)
I simply make new product configurations, put the feature that contains my application plugin and then exporting product I have this problem.
My eclipse version is 3.6.1
I think the problem is in the launcher configuration but I don't know what I have to change
In theory, your exported RCP application should simply pick up any new features and plugins you add later. However, make sure you put the extra stuff in the dropins folder of your RCP, since this is the expected location for manually installed features since the introduction of p2.
If for some reason it still doesn't work, try launching Eclipse with the -clean command line argument.
According to The Eclipse runtime options page, the -clean argument has the following effect:
[...] any cached data used by the OSGi framework and eclipse runtime will be wiped clean. This will clean the caches used to store bundle dependency resolution and eclipse extension registry data. Using this option will force eclipse to reinitialize these caches.
In my experience this can resolve many mysterious cases, where newly added plugins are not loaded by Equinox.
I am currently trying to get my headless pde-build working but I am stuck on a point where I do not know how to continue.
The problem is how to define the related target platform to compile the plugins against.
I have a build.bat with the following call (all in one line!):
java -jar D:\target\eclipse\plugins\org.eclipse.equinox.launcher_1.0.201.R35x_v20090715.jar
-application org.eclipse.ant.core.antRunner
-f D:\target\eclipse\plugins\org.eclipse.pde.build_3.5.2.R35x_20100114\scripts\productBuild\productBuild.xml
-Dbuilder=c:\pde-build\scripts %*
I tried to create the target eclipse platform from different parts: The eclipse SDK, RCP SDK, Delta Pack, PDE-SDK in all combinations but none of them worked well.
I got the following error:
BUILD FAILED
D:\target\eclipse\plugins\org.eclipse.pde.build_3.5.2.R35x_20100114\scripts\productBuild\productBuild.xml:18: Cannot fin
d ${eclipse.pdebuild.scripts}/build.xml imported from D:\target\eclipse\plugins\org.eclipse.pde.build_3.5.2.R35x_2010011
4\scripts\productBuild\productBuild.xml
where the variable ${eclipse.pdebuild.scripts} does not got resolved. I also tried to give this parameter via the command line but then I got another error regarding missing svn task which is absolutely confusing as this is working with my local eclipse installation referenced.
When I replace the path from d:/target/eclipse to my local eclipse installation the pde build works as expected!
This leads my to the point that the configuration of the target eclipse is not correct but in the moment I have no idea how to configure this!
My goal is the automate the pde build first on my local site without referencing my local eclipse and later on integrate this building process into our running cruisecontrol instance.
As I saw already another question about defining the target eclipse I would be happy if anyone can contribute hints or facts regarding the problem.
Regards,
Andreas
When performing a headless build, the target can be separate from the eclipse that is actually running the build itself. The problem you had here is that the eclipse that you were using to run the build did not have PDE/Build properly installed.
This is why the ${eclipse.pdebuild.scripts} was not set, because PDE/Build was not installed into that eclipse instance, the org.eclipse.pde.build bundle was not resolved and the code that sets this property never got called. Similarly, the necessary ant classpath entries for PDE/Build tasks would not have been set up properly either.
You need the Eclipse with PDE installed inside to run the build, but the target for the build can be separate from this.
In the build.properties file found under -Dbuilder=c:\pde-build\scripts you can set several properties:
baseLocation This is a path to an eclipse that is your target.
buildDirectory This is where the build will actually take place, source is fetched to plugins/ and features/ subfolders, but if there are already binary plugins located here then those become part of the target as well.
pluginPath This is a list of paths (separated with ';' on windows or ':' on linux) containing other locations that should be considered as part of your target. These locations can be several things:
The root of an eclipse-like install with plugins/ and features/ subfolders. This is a good way to provide the delta-pack instead of just unzipping it on top of an eclipse install.
The root of a workspace-like folder, where all subfolders are treated as plugins or features depending on the presence of a manifest or feature.xml.
The root of a bundle or feature, or the jar for a bundle.
If you are doing a p2 build (p2.gathering = true) you can also provide p2 repositories under a ${repoBaseLocation} which will be transformed and placed under ${transformedRepoLocation} and will become part of your target, and the p2 metadata there will get reused during the build.
after some more time of investigation I found out, what I did wrong so far. As I mentioned above defining the target platform is not that easy as copying the SDK and plugins in into one location (as it was in early times of eclipse dev).
The working solution by now is the following: Copying the eclipse SDK into the target location and run this version. Install inside this the neccessary PDE-Tools to enable plugin development. After that, close the IDE and copy the delta pack + the respective svn plugin (I used org.eclipse.pde.build.svn-1.0.1RC2 from sourceforge) into the target platform and you're done.
Now my automated PDE build is running as expected.
Only minor issue now is the following: The result product contains eclipse-specific menu entries which are not there when I ran this from inside my dev-eclipse.
Any hints on that?
I just posted an answer to my question on this kind of topics, may be this can help you:
Plugin product VS Feature product
I want to create an automated installer for an Eclipse plugin (i.e. not through the "Update Manager"). My scenario is simple: the user closes Eclipse, drops a downloaded JAR into the dropins folder, starts Eclipse and the rest of the process is automated.
In older Eclipse versions, before the era of P2, Eclipse had (still has) a class called InstallCommand which could be used to install pluings into the currently running platform.
While this still works in Eclipse 3.4 & 3.5, it is not behaving properly: most noticeably, plugins installed that way cannot be automatically uninstalled (it is dimmed).
The JavaDoc claims the InstallCommand is deprecated and should be replaced by a P2 alternative. However, I couldn't find the right tool for the job. There is the P2 director, but it is built for running as a separate application from the command line. It is possible to invoke it from within Eclipse but it is really not cut out for that. For example, progress monitoring and error reporting are not working well.
Does anybody know of a good alternative for that?
Thanks,
Zviki
Dropins seems very close to what you want, especially if they are just downloading jars without the associated metadata (ie the metadata will need to be auto-generated).
You could consider defining a second dropins area to manage yourself. Take a look at ProfileSynchronizer in org.eclipse.equinox.p2.reconciler.dropins, in particular the method createProfileChangeRequest. I expect the uninstall behaviour you don't like is a result of the IInstallableUnit.PROP_PROFILE_LOCKED_IU property being added.
The dropins are reconciled at startup, see the p2.reconciler.dropins Activator.watchDropins(), you can likely do the same from your own bundle to watch another folder.
I suggest to deploy your plugin as an executable JAR. The installer in the JAR should ask for the Eclipse install directory and unpack the plugin in the right place (plus some more checks as needed).
Optionally include a little "watchdog" plugin which doesn't depend on much and just checks that your main plugin loads correctly and displays a useful error message which the user can email to you for support.
According to information in bug 311590 1 which is referenced in the deprecation comment of InstallCommand an alternative is possibly to use P2 operations 2, 3.