GitHub Integration - Do Not Commit Deletes

Hi there,

I'm a big fan of the GitHub integration for PA Cloud. We use it regularly to promote changes through our environments. I believe there is an error in the way commits happen through the PA Cloud interface. I would like a commit from the IBM Planning Analytics workbench interface to NOT commit deletes to GitHub. If the commit DOES NOT commit deletes, more native GitHub features, like pull requests, will work correctly.

Consider this use case.

1. There are three environments (dev, qa, test).

2. A developer changes a turbo integrator (ti) process in the dev environment.

3. The developer is now happy with their updated ti and want to promote it by pushing it to a new feature branch (main-mynewfeature) based on the main branch.

4. The developer uses the PA Cloud Workbench to push their new ti to the GitHub repository in this new main-mynewfeature branch. main-mynewfeature is a branch of main created in the workbench user interface when the developer does this 'Add objects to repository' action.

5. Now that the new code has been pushed, the QA engineer wants to merge this change into the main branch. By using a pull request in GitHub, they review the changes that would be applied as part of the pull request for main<-main-mynewfeature

6. The QA engineer sees that this pull request plans to delete all files in the main branch, except for the ones explicitly in the feature branch

7. The QA engineer cannot approve this pull request because it would delete important files, that should not change, as part of this pull request.

Observation - Excluded Files Are Committed as Deletes

When the developer pushes their update to a new feature branch (main-mynewfeature) the user interface gives them an option to select which objects should be included. This is correct. The developer can pick the objects that have changed as part of this new feature. The commit to the feature branch should be ONLY added/changed objects. The commit is registering that ALL other files are to be deleted as part of a pull request. This is not correct.

Observation - PA Cloud Workbench Works Correctly for Feature Branches

Given the same use case above, a developer can connect to workbench in production and use the 'Update model from repository' feature to pull just the changed objects in the feature branch into the target environment. This is correct. The objects excluded from the feature branch ARE NOT DELETED.

At the moment, we cannot use the pull request functionality in GitHub because every pull request will try to delete all files excluded from the feature branch. Lack of source code control has always been a criticism of TM1/Planning Analytics. Improving these hooks into GitHub can demonstrate that we have mature practices around source code control, and open the door for continuous integrationa and continuous delivery (ci/cd) automation.