This is more of a code organization question (in Go in this case) - is there a common pattern for how to organize the Pulumi code? Specifically, I’m managing quite a few Kubernetes resources, so everything is quite verbose, probably a couple of thousand lines of code. Typically I’d break different parts into functions (and maybe eventually a component resource), but they also have dependencies between them. As such, I’d have to ensure my functions return the created resources (which may be several in “function”) and also take in all the inputs they need. That’s a lot of overhead relative to just putting everything “inline/procedurally” in a single function. I’m just curious how folks manage this in terms of returning resources and passing them in dependencies between different parts.

We're a few weeks into our migration over to Pulumi, and we're using a similar pattern in Golang. One thing we did was break out our infra into multiple Pulumi projects, and then use

StackReference

to import resources across project boundaries. To manage resource dependencies inside the same project, we're defining higher level

struct

types to encapsulate and group resources logically, and pass those structs around. This keeps our functions simpler and prevents parameter bloat.

Thanks @green-musician-49057 - good to hear from other experiences. Using

StackReference

makes sense for other stacks, and no issue with that, it’s more the second case you mentioned within the same project/stack, just how to manage the various resources.

👍