Is there a way to generate a new cloudformation template within the Deploy lifecycle hook?

[taphill]

Hello All! I have a question that I'm hoping someone could answer for me.

Here is my use case:

My company has a project with some @http and @scheduled functions. We would like to deploy all the @http functions to both the staging and production environments, but we only want the @scheduled functions deployed in the production environment.

Currently, I'm able to accomplish this like this:

module.exports = {
  deploy: {
    start: async ({ arc, cloudformation, dryRun, inventory, stage }) => {
      if (stage !== "staging") return;

      delete cloudformation.Resources.FunctionOneScheduledLambda
      delete cloudformation.Resources.FunctionOneScheduledEvent
      delete cloudformation.Resources.FunctionOneScheduledPermission

      // etc...

      return cloudformation;
    },
  },
};

This works fine but is slightly tedious to do for every function and a little prone to user error if someone forgets to edit this file when they add a new function, or if the function name changes for some reason.

My question is this:

Is it possible to edit either the arc object or the inventory object and then use that to generate a new cloudformation template? Ideally, I would be able to do something like this:

module.exports = {
  deploy: {
    start: async ({ arc, cloudformation, dryRun, inventory, stage }) => {
      if (stage !== "staging") return;

      const newInventory = {
        ...inventory,
        inv: {
          ...inventory.inv,
          scheduled: null,
        },
      };

       // Is there something like this?
      const newCloudformation = createCloudformation(newInventory);

      return newCloudformation;
    },
  },
};

I looked through the docs, and some examples of other plugins, but I can't seem to find something similar.

Future possibilities:

If this was possible it could enable a pretty cool plugin that would allow you to ignore deploying certain functions based on the environment which may be useful. Something like this...just thinking out loud!

@deploy-ignore
staging
  # functions to ignore...

productions
  # functions to ignore...

[ryanblock]

The Arc and Inventory objects are intentionally deep-frozen, so unfortunately that path is not available to you. My suggestion would be to create your @scheduled functions conditionally via the set.scheduled API. Return nothing in !production, and return the stuff you want to deploy in prod.

One note: Arc's set APIs do not have a stage parameter, so use inventory.inv._arc.deployStage to infer the intended stage. deployStage will be null when local (e.g. run in Sandbox).

[taphill]

This worked for me, thank you @ryanblock!