CUE templating advice

[artemlive]

Hello!
First of all, thanks for such a great tool. I'm enjoying playing around with Timoni.
I'm trying to learn how to work with CUE, and I want to ask for advice. I want to create an abstract package that can create various configmaps/services/deployments, etc.
In my experiments, I have values like this:

values: {
	configMaps: "first": {
		annotations: {
			"test":  "testsstestsssssstestssssssssss"
			"test2": "test2"
		}
		name: "first"
		data: {"1": ""}
	}
	configMaps: "second": {
		annotations: {
			"foo": "bar"
		}
		name: "second"
		data: {
			"1": "2"
			"2": """
                               test
                               test2
                               """
		}
	}
}

And then in the instance:

#CMConfig: {
	// Metadata (common to all resources)
	annotations?: {[string & =~"^(([A-Za-z0-9][-A-Za-z0-9_./]*)?[A-Za-z0-9])?$" & strings.MaxRunes(63)]: string}
	name!: string & =~"^(([A-Za-z0-9][-A-Za-z0-9_.]*)?[A-Za-z0-9])?$" & strings.MaxRunes(63)
	data!: {[string]: string}
}

// Instance takes the config values and outputs the Kubernetes objects.
#Instance: {
	config: #Config
	objects: {
		for c,v in config.configMaps {
				"cm-\(c)":  #ConfigMap & {
						_config: config
						_cm_config: #CMConfig & {
							annotations: v.annotations
							name: v.name
							data: v.data
						}
				}
		}

And it works as I expected:

---
apiVersion: v1
data:
  "1": ""
immutable: true
kind: ConfigMap
metadata:
  annotations:
    test: testsstestsssssstestssssssssss
    test2: test2
  labels:
    app.kubernetes.io/name: backend
    app.kubernetes.io/version: 0.0.0-devel
  name: first
  namespace: default
---
apiVersion: v1
data:
  "1": "2"
  "2": |-
    test
    test2
immutable: true
kind: ConfigMap
metadata:
  annotations:
    foo: bar
  labels:
    app.kubernetes.io/name: backend
    app.kubernetes.io/version: 0.0.0-devel
  name: second
  namespace: default

But I'm curious: am I using the right approach? There are only a few examples of how to work with Timoni, and they are straightforward when values are passed to the statically defined objects. So I wonder whether I'm doing something wrong or it looks fine. :)
In perspective, I want to migrate all our helm charts to Timoni.
Thanks for any advice.

[stefanprodan]

Looking at the output I guess your #ConfigMap does uses the corev1 schema and the Timoni #Metadata. One thing I would change is not repeat the name, but use the map key for it.

		for c,v in config.configMaps {
				"cm-\(c)":  #ConfigMap & {
						_config: config
						_cm_config: #CMConfig & {
							annotations: v.annotations
							name: c
							data: v.data
						}
				}
		}

I would also set max length for the data keys in data!: {[string]: string}, and maybe a regex if these keys end up as files inside pods, or env vars.

[artemlive]

Thanks for the suggestions. I'll make the changes while playing with the syntax. So, is the approach itself idiomatically correct?
It's a little bit difficult to switch from the helm go templates syntax to the complexity of cuelang :)

[stefanprodan]

It's definitely more complex than Go text templating, it took me a week or two to get used to it. The most tricky part, coming from Helm and Kustomize, was getting used to CUE's immutability as it requires a diffrent way of thinking about cfg management. For example, once you set a field to a concrete value e.g. replicas: 1, you can no longer change it. At first, I've seen this as major blocker, but later on I realise this is actually a sane way to deal with configs. It forces you to define inputs with a strict schema and defaults. You no longer have to dig into layers of mutations to figure out why some field ended up with a particular value.

[artemlive]

Thanks for your reply.
It inspires me to gain a more profound knowledge of Timoni and CUE.