This is actually the expected behavior, and not at all an anomaly The reasoning behind this is that we want to allow the user to deploy partially executed workflows, and then only resume execution from where the workflow was saved. E.g., you can upload a workflow in which data has already been read and processed, so that on the WebPortal, only the visualization has to be processed. This can significantly speed up the time it takes for the visualizations to be shown.
Obviously, this has the drawback that a fully executed workflow will not show anything. Still, this is hard to catch upon uploading the workflow, since at that point, we don’t actually know whether the workflow will run on the WebPortal or be called from the client / via REST. If in doubt, the “Deploy to Server” dialog does offer the option to reset before upload, which is a reasonable thing to do unless you have good reasons why you want to upload the workflow in executed state.