update docs

Author: ArturWierzbicki

docs/resources/apps_generic_resource.md

@@ -3,12 +3,11 @@
 page_title: "grafana_apps_generic_resource Resource - terraform-provider-grafana"
 subcategory: "Grafana Apps"
 description: |-
-  Manages arbitrary Grafana App Platform resources when a typed Terraform resource is not yet available.
+  Manages arbitrary Grafana App Platform resources when a typed Terraform resource is not yet available. The resource is still experimental; diffing semantics are subject to change - feedback welcome in https://github.com/grafana/terraform-provider-grafana/issues.
   This resource accepts a Kubernetes-style manifest as the single source of truth for the resource definition. Use HCL merge() if you need to inject Terraform variables into a static manifest file.
-  Only namespaced App Platform kinds are supported. The provider autodiscovers the namespace from /bootdata on every operation. If autodiscovery does not find a cloud stack namespace, the provider falls back to the explicit stack_id and then org_id provider settings. This means cloud instances work correctly even when org_id is set for legacy API compatibility.
-  Top-level manifest fields are limited to apiVersion, kind, metadata, spec, and the ignored status field. That allowlist is only for the top level of manifest; manifest.metadata itself is not restricted to a fixed field list. If metadata.namespace is configured, it must match the provider-selected namespace.
+  Only namespaced App Platform kinds are supported. The provider autodiscovers the namespace from /bootdata on every operation. If autodiscovery does not find a cloud stack namespace, the provider falls back to the explicit stack_id and then org_id provider settings.
+  Top-level manifest fields are limited to apiVersion, kind, metadata, spec, and the ignored status field. If metadata.namespace is configured, it must match the provider-selected namespace.
   Inside manifest.metadata, both Kubernetes name and uid are accepted as input aliases for the object identifier.
-  The provider discovers the plural API route from Grafana's App Platform discovery endpoints for every operation. If discovery cannot resolve a namespaced route for the requested kind, the operation fails.
   The top-level secure argument is write-only and requires Terraform 1.11 or later. Each configured key must set exactly one of create or name, and Terraform only re-sends those secure values when secure_version changes.
   Reads refresh managed drift from the API. Metadata drift is limited to the metadata keys you configured; spec is authoritative, so extra remote spec fields are refreshed into state and will drift until Terraform restores the configured object.
   Import format:
@@ -20,18 +19,16 @@ description: |-
 
 # grafana_apps_generic_resource (Resource)
 
-Manages arbitrary Grafana App Platform resources when a typed Terraform resource is not yet available.
+Manages arbitrary Grafana App Platform resources when a typed Terraform resource is not yet available. The resource is still experimental; diffing semantics are subject to change - feedback welcome in https://github.com/grafana/terraform-provider-grafana/issues.
 
 This resource accepts a Kubernetes-style `manifest` as the single source of truth for the resource definition. Use HCL `merge()` if you need to inject Terraform variables into a static manifest file.
 
-Only namespaced App Platform kinds are supported. The provider autodiscovers the namespace from `/bootdata` on every operation. If autodiscovery does not find a cloud stack namespace, the provider falls back to the explicit `stack_id` and then `org_id` provider settings. This means cloud instances work correctly even when `org_id` is set for legacy API compatibility.
+Only namespaced App Platform kinds are supported. The provider autodiscovers the namespace from `/bootdata` on every operation. If autodiscovery does not find a cloud stack namespace, the provider falls back to the explicit `stack_id` and then `org_id` provider settings.
 
-Top-level manifest fields are limited to `apiVersion`, `kind`, `metadata`, `spec`, and the ignored `status` field. That allowlist is only for the top level of `manifest`; `manifest.metadata` itself is not restricted to a fixed field list. If `metadata.namespace` is configured, it must match the provider-selected namespace.
+Top-level manifest fields are limited to `apiVersion`, `kind`, `metadata`, `spec`, and the ignored `status` field. If `metadata.namespace` is configured, it must match the provider-selected namespace.
 
 Inside `manifest.metadata`, both Kubernetes `name` and `uid` are accepted as input aliases for the object identifier.
 
-The provider discovers the plural API route from Grafana's App Platform discovery endpoints for every operation. If discovery cannot resolve a namespaced route for the requested kind, the operation fails.
-
 The top-level `secure` argument is write-only and requires Terraform 1.11 or later. Each configured key must set exactly one of `create` or `name`, and Terraform only re-sends those secure values when `secure_version` changes.
 
 Reads refresh managed drift from the API. Metadata drift is limited to the metadata keys you configured; `spec` is authoritative, so extra remote spec fields are refreshed into state and will drift until Terraform restores the configured object.

internal/resources/appplatform/generic/generic_resource.go

@@ -169,18 +169,16 @@ func (r *genericResource) Schema(_ context.Context, _ tfrsc.SchemaRequest, resp
 	resp.Schema = schema.Schema{
 		Description: "Manages arbitrary Grafana App Platform resources when a typed Terraform resource is not yet available.",
 		MarkdownDescription: `
-Manages arbitrary Grafana App Platform resources when a typed Terraform resource is not yet available.
+Manages arbitrary Grafana App Platform resources when a typed Terraform resource is not yet available. The resource is still experimental; diffing semantics are subject to change - feedback welcome in https://github.com/grafana/terraform-provider-grafana/issues.
 
 This resource accepts a Kubernetes-style ` + "`manifest`" + ` as the single source of truth for the resource definition. Use HCL ` + "`merge()`" + ` if you need to inject Terraform variables into a static manifest file.
 
-Only namespaced App Platform kinds are supported. The provider autodiscovers the namespace from ` + "`/bootdata`" + ` on every operation. If autodiscovery does not find a cloud stack namespace, the provider falls back to the explicit ` + "`stack_id`" + ` and then ` + "`org_id`" + ` provider settings. This means cloud instances work correctly even when ` + "`org_id`" + ` is set for legacy API compatibility.
+Only namespaced App Platform kinds are supported. The provider autodiscovers the namespace from ` + "`/bootdata`" + ` on every operation. If autodiscovery does not find a cloud stack namespace, the provider falls back to the explicit ` + "`stack_id`" + ` and then ` + "`org_id`" + ` provider settings.
 
-Top-level manifest fields are limited to ` + "`apiVersion`" + `, ` + "`kind`" + `, ` + "`metadata`" + `, ` + "`spec`" + `, and the ignored ` + "`status`" + ` field. That allowlist is only for the top level of ` + "`manifest`" + `; ` + "`manifest.metadata`" + ` itself is not restricted to a fixed field list. If ` + "`metadata.namespace`" + ` is configured, it must match the provider-selected namespace.
+Top-level manifest fields are limited to ` + "`apiVersion`" + `, ` + "`kind`" + `, ` + "`metadata`" + `, ` + "`spec`" + `, and the ignored ` + "`status`" + ` field. If ` + "`metadata.namespace`" + ` is configured, it must match the provider-selected namespace.
 
 Inside ` + "`manifest.metadata`" + `, both Kubernetes ` + "`name`" + ` and ` + "`uid`" + ` are accepted as input aliases for the object identifier.
 
-The provider discovers the plural API route from Grafana's App Platform discovery endpoints for every operation. If discovery cannot resolve a namespaced route for the requested kind, the operation fails.
-
 The top-level ` + "`secure`" + ` argument is write-only and requires Terraform 1.11 or later. Each configured key must set exactly one of ` + "`create`" + ` or ` + "`name`" + `, and Terraform only re-sends those secure values when ` + "`secure_version`" + ` changes.
 
 Reads refresh managed drift from the API. Metadata drift is limited to the metadata keys you configured; ` + "`spec`" + ` is authoritative, so extra remote spec fields are refreshed into state and will drift until Terraform restores the configured object.