GCP.CloudFunctions.Function deploys the same Effect program as
Cloud Run, built a different way: Alchemy
bundles your code for Node.js, uploads it, and Google builds it. You
can deploy from a machine without Docker.
Declare a function
// src/Notes.ts
import * as GCP from "@alchemy.run/gcp";
import * as Effect from "effect/Effect";
import * as HttpServerResponse from "effect/http/HttpServerResponse";
import { NotesDb } from "./resources.ts";
export default class Notes extends GCP.CloudFunctions.Function<Notes>()(
"Notes",
{ main: import.meta.url },
Effect.gen(function* () {
const db = yield* GCP.Firestore.ReadWriteDatabase(NotesDb);
return {
fetch: Effect.gen(function* () {
const { documents } = yield* db.list("notes", { pageSize: 100 });
return yield* HttpServerResponse.json({
notes: documents.map((doc) => doc.fields),
});
}).pipe(Effect.orDie),
};
}).pipe(Effect.provide(GCP.Firestore.ReadWriteDatabaseHttp)),
) {} The function runs on Node.js 22 and serves fetch through Google's
Functions Framework.
Add it to the Stack
// alchemy.run.ts
Effect.gen(function* () {
const notes = yield* Notes;
return { url: notes.url };
}), Make it public
A 2nd-gen function is served by a Cloud Run service. Grant
roles/run.invoker to allUsers on that service to open it up:
yield* GCP.IAM.Member("PublicInvoker", {
kind: "run.service",
name: notes.service.as<string>(),
role: "roles/run.invoker",
member: "allUsers",
}); Without it, callers need a Google identity token.
Consume a topic
Event sources work as on Cloud Run. Provide the Cloud Functions implementation:
yield* GCP.PubSub.consumeTopicMessages(Orders, (messages) =>
messages.pipe(Stream.runForEach(({ message }) => Effect.log(message))),
);
// …provided with Effect.provide(GCP.CloudFunctions.TopicEventSource) Deploy your own source
Leave out main to deploy an archive you built yourself:
const hello = yield* GCP.CloudFunctions.Function("Hello", {
buildConfig: {
runtime: "nodejs22",
entryPoint: "helloHttp",
source: { storageSource: { bucket: "my-source", object: "hello.zip" } },
},
}); Cloud Functions or Cloud Run?
Both run your program on Cloud Run. Pick Cloud Functions when you can't run Docker where you deploy. Pick a Cloud Run service otherwise: it deploys faster and gives you every revision setting.
Example
examples/gcp-cloud-function
is a notes API over Firestore.
Where next
- Cloud Run — the same program with a locally built image.
- CloudFunctions reference — every prop and attribute.