Ricky's Hodgepodge

These days I have been involved in several issues about the old Java Server Faces (JSF) technology. Currently wildfly is using the former Oracle driven mojarra implementation that, as the rest of the JavaEE specification, now is controlled by the eclipse foundation. Working with the code the major issue I have faced has been launching the tests. Although it is explained in the project, I spent some time to fully understand the process, so better if I collect the commands here.

1. As the documentation says the glassfish javaee server is needed, but the version is specially important when testing the different branches. For example if you want to test branch 2.3 (2.3.14 right now) you can install glassfish 5.0, but if you want to test the master branch (future 3.0 version) you need glassfish 5.1 from the same eclipse project.

   Let's assume we want to test the master branch. Then the glassfish 5.1 should be installed in your computer.

   
   unzip glassfish-5.1.0.zip
   mv glassfish5 glassfish5.1
   

2. Now just clone and compile the project (default maven usage).

   
   mvn clone https://github.com/eclipse-ee4j/mojarra.git
   cd mojarra
   mvn clean install
   

3. In order to launch a specific app test the documentation can just be followed, using the properties to locate the glassfish instance. The test application test/javaee7/cdiInitDestroyEvent is used as example.

   
   cd test/javaee7/cdiInitDestroyEvent
   

   • Copy the compiled JSF jar file into the glassfish.

     
     mvn -Pglassfish-patch -Dglassfish.cargo.home=/home/rmartinc/glassfish5.1 -Dglassfish.patch.home=/home/rmartinc/glassfish5.1 validate
     

   • Start the domain with thew new JSF version.

     
     mvn -Pglassfish-cargo -Dglassfish.cargo.home=/home/rmartinc/glassfish5.1 -Dglassfish.patch.home=/home/rmartinc/glassfish5.1 cargo:start
     

   • Deploy the application for the test.

     
     mvn -Pglassfish-cargo -Dglassfish.cargo.home=/home/rmartinc/glassfish5.1 -Dglassfish.patch.home=/home/rmartinc/glassfish5.1 cargo:redeploy
     

   • Launch the tests against the app just deployed.

     
     mvn -Pintegration -Dglassfish.cargo.home=/home/rmartinc/glassfish5.1 -Dglassfish.patch.home=/home/rmartinc/glassfish5.1 verify
     

   • If everything goes OK you can stop the server.

     
     mvn -Pglassfish-cargo -Dglassfish.cargo.home=/home/rmartinc/glassfish5.1 -Dglassfish.patch.home=/home/rmartinc/glassfish5.1 cargo:stop
     

4. The documentation also comments that all the tests inside a directory can be executed. To do that the same steps are done but in the upper directory. For example the javaee7 directory is used to run all the tests in that version of the JavaEE standard (the -fn parameter is passed to not stop tests on error, I want all of them executed even if there are errors):

   
   cd test/javaee7
   mvn -Pglassfish-patch -Dglassfish.cargo.home=/home/rmartinc/glassfish5.1 -Dglassfish.patch.home=/home/rmartinc/glassfish5.1 validate
   mvn -Pglassfish-cargo -Dglassfish.cargo.home=/home/rmartinc/glassfish5.1 -Dglassfish.patch.home=/home/rmartinc/glassfish5.1 cargo:start
   mvn -Pglassfish-cargo -Dglassfish.cargo.home=/home/rmartinc/glassfish5.1 -Dglassfish.patch.home=/home/rmartinc/glassfish5.1 cargo:redeploy
   mvn -fn -Pintegration -Dglassfish.cargo.home=/home/rmartinc/glassfish5.1 -Dglassfish.patch.home=/home/rmartinc/glassfish5.1 verify
   

5. Finally there are some test directories that do not run OK. But then I realized that, in the automatic tests that are launched when a Pull Request is received, only the following folders are executed: htmlunitaware, javaee6, javaee6web, javaee7, javaee8, servlet31, servlet40. I can confirm that servlet30 also runs without any problem but all the others report an issue at some point (I think just because they are not used and therefore not maintained, except the cluster folder, that needs several instances and it is not explained how to test it, maybe it just needs a glassfish cluster deployed in some specific ports but I did not check it deeply).

And that's all. Very simple entry. The documentation also comments that the tests can be run with different options (different JSF configurations). Indeed my main problem was knowing what glassfish to use for the tests, because I initially tried with 5.0 and that does not work for the master branch (you need the new version 5.1 to be compatible). Maybe I'm just repeating the same statements that there are in the documentation, but I prefer to write a summary here with the exact commands I used, just in case I need to work again on the mojarra implementation.

Happy testing!

Today's entry is a short post about using Wildfly and OpenSSL in a Windows system. I needed to work on the wildfly-openssl library, which provides to Wildfly the ability of using OpenSSL in order to implement the TLS/SSL protocols instead of the default java JSSE infrastructure. The OpenSSL library is ubiquitous inside the linux world but rarely used in Windows. In the Microsoft OS some custom steps are needed to integrate it in the Wildfly server and, although they are not very complicated, better if I summarize the procedure here. So the entry is just that, how to configure Wildfly to use the OpenSSL provider in a Windows 2016 box.

1. The first step is installing the OpenSSL software in the box. For that I used this package which I reached from the OpenSSL site itself. I really do not know if this specific bundle is used or not, I just needed some quick binaries to play with them. The chosen version 1.1.0L needs the windows C++ runtime 2013. So first the re-distributable package should be installed.

2. Once the C++ libraries are in the system just continue with the OpenSSL. The software was installed into the directory C:\Program Files\OpenSSL-Win64. The DDL library files were copied to the bin folder instead of the system32 system location. At this point we can check if the OpenSSL works (just the version option is needed to see if everything is installed OK).

   
   openssl.exe version
   OpenSSL 1.1.0l  10 Sep 2019
   

3. A sample self-signed certificate is created to be used inside the server.

   
   cd %JBOSS_HOME%\bin
   keytool -genkeypair -alias localhost -keyalg RSA -keysize 1024 -validity 365 -keystore ..\standalone\configuration\keystore.jks -dname "CN=localhost" -storepass XXXX
   

4. Now the Wildfly standalone server should be started passing the path to the OpenSSL librariles (as the DDL files were installed in an unusual directory, both libraries are directly passed as java startup options).

   
   standalone.bat -b 0.0.0.0 -Dorg.wildfly.openssl.path.ssl="C:\\Program Files\\OpenSSL-Win64\\bin\\libssl-1_1-x64.dll" -Dorg.wildfly.openssl.path.crypto="C:\\Program Files\\OpenSSL-Win64\\bin\\libcrypto-1_1-x64.dll"
   

5. Finally the standalone server is configured to use the openssl and that certificate with the following CLI commands.

   
   /subsystem=elytron/key-store=sslKS:add(path=keystore.jks, relative-to=jboss.server.config.dir, credential-reference={clear-text=XXXX}, type=JKS)
   /subsystem=elytron/key-manager=sslKM:add(key-store=sslKS, algorithm="SunX509", credential-reference={clear-text=XXXX})
   /subsystem=elytron/server-ssl-context=sslSSC:add(providers=openssl, key-manager=sslKM, protocols=["TLSv1.2"])
   /subsystem=elytron/server-ssl-context=sslSSC:write-attribute(name=providers, value=openssl)
   batch
   /subsystem=undertow/server=default-server/https-listener=https:undefine-attribute(name=security-realm)
   /subsystem=undertow/server=default-server/https-listener=https:write-attribute(name=ssl-context, value=sslSSC)
   run-batch
   reload
   

And that is all. Now the Wildfly can be accessed at secure port 8443 using the OpenSSL library inside the windows 2016 machine. Time to start what I really wanted to do.

Regards.

Another short entry to announce that these weeks I have been working in adding tags to my runnerup backend server. I missed the idea of tagging the activities with meaningful labels that I can use later to search of filter them. That feature has just been added. Remember that runnerupweb is just a toy project that I mainly use to do a backup of my running activities, do not think that it is a full application. My first idea was simply storing the TCX files into my nextcloud server, just that, but, as I already had something developed before, I continued with the idea in order to visualize and get some metrics from the workouts (once runalyze was closed).

The main idea about the tags is that activities can be labeled manually (the activity is marked by the user using the app) or automatically (when the TCX file is uploaded). So now the user can create tags that can be assigned to the workouts. The automatic tags are assigned to the training activity without user interaction and they are created receiving an activity as a sample (that means that the automatic tag takes an activity to obtain some info from it, then the tag can check if another one is assignable or not). The current automatic tags inside the app are the following:

• Sport: This tag just checks the sport type of the activity and assign a label to it (running, biking,...).

• Location: This tag takes the first track-point of the TCX file and checks if the activity is inside a specific radius (I wanted this because sometimes I run in several cities and I wanted to automatically tag the activities with the city).

• Regular expression: Simple tag that uses a PHP regexp over the notes of the TCX file. For example if a tag is created using the expression /race/i any activity uploaded with the word race inside the notes will be tagged with this label.

• Route: This is the more complex tag and really the goal I wanted to achieve. I usually do different routes when running (mainly three) and I wanted to automatically tag the TCX file with the route done. The idea is the tag uses the sample activity to get N points and then calculate the distance between the original points and the new activity points, if the accumulated distance is less than a tolerance error (indeed a percent of the distance), the new activity is labeled with the route tag.

• Speed and Distance: Another two simple tags that just tag the new activity based on the average speed (for example less than 5 minutes per kilometer) or the total distance of the training (more than 10 km).

Here I present some pictures of how the tags look.

• List of user tags:

  

• Using tag filter in the activity list:

  

• Tags inside the activity:

  

Now I have to upgrade my desktop (real) application. Create my tags and then re-calculate all my files to automatically assign the labels to them. Let's see how it works. If I see no big problems a new tag 0.2.0 will be created for the project.

Regards.