You wouldn't want a provider that can deploy to more than one region or whatever. That would allow for accidentally deploying to the wrong place. If one function's side effect changes the region.. chaos could ensue.

am I correct in thinking a “Provider” is basically some settings on the AWS client?

auth + region, that kind of thing

doesn’t really control any business logic

Hey @little-cartoon-10569 — after using a customer provider, I’m seeing that since the provider is “changing”, it tries to recreate the resource. Do you know of a way around this?

When changing a provider, the resources do get destroyed, unfortunately. There are workarounds, none particularly pleasant.

You can create the new provider with the same settings as the default provider,, then export the stack, edit all the resources to use the explicit provider, and import the stack.

You can create a new stack that uses the explicit provider, import all the resources to it, and delete them from the old stack.