One of the most frustrating aspects of working with HCI is the buggy development environment of the Eclipse-based Integration Designer plugin.
While in general some of bugs are just minor annoyances (error pop ups that can be “cured” by a restart), the most disastrous is when a perfectly working iFlow no longer works after some changes. Even though the processing logic of the iFlow is reverted back to the original, it sometimes will still not work. Sometimes, iFlow changes will cause local checks to fail with non-descriptive errors. Sometimes, the changes will pass local checks but will fail during deployment to the tenant, again with non-descriptive errors in the tail log. It took a while to understand what was happening as it occurs in a random fashion.
As I was new to HCI, I tended to develop the iFlows in a stage-by-stage manner – include certain functionality, save, deploy and test it out before enhancing it with further functionality. Initially, the errors cause me to think that some of the logic that I was trying to implement were not possible or not supported, even though the errors do not specifically mention that. However, I found out that if I were to develop the similar logic all at once in a new iFlow, it would work!!
This led me to the conclusion that “iFlow files can be corrupted when making changes!“
In my experience, corruption tends to happen more often when there are deletion of objects or connectors in the iFlow. Although iFlows are designed using a graphical layout editor, the content is actually saved as an XML file. Therefore my guess is that when some editing is done via the graphical iFlow editor, the underlying XML is not adjusted correctly, causing corruption of the generated file.
Furthermore, HCI does not come with native version management, therefore changes are not easily reversible.
After much frustration getting my iFlows corrupted from time to time, and having to rebuilt them from scratch, I decided to use Git for version management of my HCI Integration Projects. In this blog, I will share with you how the EGit plugin within Eclipse can be used to configure the Git version management system for local development of HCI iFlows. This is similar to the approach I’ve blogged about below on using EGit for Java development in PI/NWDS.
Below are component versions of the Eclipse plugins that I was using. Hopefully future versions would provide a more stable iFlow editor.
Eclipse Plugin Versions: Adapter 2.11.1, Designer 2.11.1, Operations 2.10.0
In order to use EGit, the EGit plugin needs to be installed in the Eclipse environment. For those using Luna SR2 (4.4.2), it should already come together with EGit 3.4.2 as shown below.
Before using EGit in Eclipse, some initial configuration needs to be performed first. Please follow the same steps as indicated in Initial Configuration section of the following blog.
Manage Integration Project with EGit in Eclipse
Following are the steps to put an HCI Integration Project under EGit’s version management.
Step 1 – Create Git Repository
Change to Git perspective – Window > Open Perspective > Other … > Git
Click the Create a New Git Repository button.
Specify the location for the new Git repository.
Once the repository is created, it will be listed under the Git Repositories tab.
Step 2 – Import Integration Project into Git Repository
Next we want to import/share the HCI Integration Projects in the newly created Git Repository. Switch back to the Integration Designer perspective.
Right click on the project, and select Team > Share Project.
Select Git from the sharing type.
Select the recently created repository from the dropdown list for Repository.
Step 3 – Commit contents to be tracked
After the project has been shared in the Git repository, we can proceed to commit the contents that are to be tracked.
Right click on the project, and select Team > Commit.
Provide a message for the commit and select the files that are to be tracked – in general I will select all source codes (iFlow, XSD, WSDL, Scripts) and project configuration files.
Now that the contents have been committed to the Git repository, I am free to modify the contents of the iFlow (or any of the commited objects) without worrying if any changes would render the iFlow useless.
Each time I’ve made changes to the iFlow and verified that it’s working, the commit step is repeated to save a version of that to the repository.
Step 4 – Reverting iFlow to previous version
Let say after some changes have been made to the iFlow and it becomes corrupted causing errors during local check or deployment, we can revert to a previous version by right clicking the iFlow object in the Project Explorer and selecting Replace With.
We can either select HEAD revision (which is normally the last committed version) or Commit… for some other version from the commit history.
As shown, by using EGit as the version management system for HCI Integration Projects, we can avoid the potential disastrous effects of changes rendering an iFlow corrupted and useless. There are also other benefits of using EGit since there is no native version management system for HCI objects that are developed locally on Eclipse. Besides iFlow, the other objects in an Eclipse HCI Integration Project can be tracked too like WSDLs, XSDs, Groovy scripts and even Message Mapping which cannot be easily “version-managed” by other methods.
I’d definitely recommend those working on HCI with Eclipse to use EGit if there is no other version management tool in place.