It feels surprising that changing a step name in a deployment process would be able to break a script, but it in fact can for us since we need to access Output variables of a previous step.
Is there another recommended way of going about doing this? Is there some other identifier we could use to get the step output that isn’t the step name?
It’s a hidden dependency, and it seems our only option is to change relevant step names to something like “Deploy to Azure (Do not change this step name in any way)”, which is ugly to read and view, albeit safer.
Thanks for getting in touch! We need a easily accessible identifier for the step. Even if it was some other value such as (obscure and fairly hidden) ID ie (step-243), you would need to rename it “Deploy to Azure (do not delete this has it has variables needed later)”. At least with a step name, people can see fairly easily which step the variable is being pulled from. If it were chosen based on the step in the order list, if people changed the order of steps or added new ones, then again the dependency is lost. Out of all of these options Step Name is generally the least changed and easiest to identify when viewing the variables.
It’s a debatable topic for sure, but as discussed above there are pro-cons to all different ways to identify. Your safety measure seems like a good way to deal with it, if people like to change step names.