Create Custom Objects

Make sure you have everything you need before proceeding:

• You understand the concepts of Protobuf.
• You have completed the introductory CosmJS tutorial.
• Go and npm are installed.
• You have finished the checkers Go blockchain exercise. If not, you can follow the tutorial here, or just clone and checkout the relevant branch (opens new window) that contains the final version.

With your checkers application ready for use, it is a good time to prepare client elements that eventually allow you to create a GUI and/or server-side scripts. Here, you will apply what you have learned about creating your own custom CosmJS interfaces.

Before you can get into working on your application directly, you need to make sure CosmJS understands your checkers module and knows how to interact with it. This generally means you need to create the Protobuf objects and clients in TypeScript and create extensions that facilitate the use of them.

Compile Protobuf

You will have to create a client folder that will contain all these new elements. If you want to keep the Go parts of your checkers project separate from the TypeScript parts, you can use another repository for the client. To keep a link between the two repositories, add the client parts as a submodule to your Go parts:

Copy $ git submodule add git@github.com:cosmos/academy-checkers-ui.git client

Replace the path with your own repository.

In effect, this creates a new client folder. This client folder makes it possible for you to easily update another repository with content generated out of your Go code.

As noted previously when you created your project with Ignite, your project is known as alice/checkers (or the alternative name you chose).

However, github.com/cosmos/academy-checkers-ui (opens new window) is a repository that already exists and uses b9lab/checkers (opens new window) in order to maintain compatibility with the checkers application (opens new window), of which it is a submodule.

Therefore, if you choose to reuse github.com/cosmos/academy-checkers-ui, you must do one of the following:

• Make sure to replace all occurrences of b9lab/checkers with alice/checkers, or whichever name you picked.
• Re-run the protoc step so that it does this replacement for you.

If you do not do one of these actions, you may encounter ambiguous errors such as:

Copy Query failed with (6): unknown query path: unknown request at QueryClient.queryUnverified (http://localhost:3000/static/js/bundle.js:21568:13) at async Object.getAllStoredGames (http://localhost:3000/static/js/bundle.js:2975:26) at async src_checkers_stargateclient__WEBPACK_IMPORTED_MODULE_0__.CheckersStargateClient.getGuiGames

Create a folder named scripts in your project root. This is where you will launch the Protobuf compilation:

Copy $ mkdir -p scripts/protoc

In the scripts folder, or in your Docker image, install a compiler:

---

Now install your additional modules:

Create the folder structure to receive the compiled files:

Copy $ mkdir -p client/src/types/generated

Check what Cosmos SDK version you are using:

Copy $ grep cosmos-sdk go.mod

This may return:

Copy github.com/cosmos/cosmos-sdk v0.45.4

Download the required files from your .proto files:

Copy $ mkdir -p proto/cosmos/base/query/v1beta1 $ curl https://raw.githubusercontent.com/cosmos/cosmos-sdk/v0.45.4/proto/cosmos/base/query/v1beta1/pagination.proto -o proto/cosmos/base/query/v1beta1/pagination.proto $ mkdir -p proto/google/api $ curl https://raw.githubusercontent.com/cosmos/cosmos-sdk/v0.45.4/third_party/proto/google/api/annotations.proto -o proto/google/api/annotations.proto $ curl https://raw.githubusercontent.com/cosmos/cosmos-sdk/v0.45.4/third_party/proto/google/api/http.proto -o proto/google/api/http.proto $ mkdir -p proto/gogoproto $ curl https://raw.githubusercontent.com/cosmos/cosmos-sdk/v0.45.4/third_party/proto/gogoproto/gogo.proto -o proto/gogoproto/gogo.proto

Now compile:

You should now have your TypeScript files.

In order to easily repeat these steps in the future, you can add them to your existing Makefile with slight modifications:

Copy install-protoc-gen-ts: mkdir -p scripts/protoc cd scripts && npm install curl -L https://github.com/protocolbuffers/protobuf/releases/download/v21.5/protoc-21.5-linux-x86_64.zip -o scripts/protoc/protoc.zip cd scripts/protoc && unzip -o protoc.zip rm scripts/protoc/protoc.zip ln -s $(pwd)/scripts/protoc/bin/protoc /usr/local/bin/protoc cosmos-version = v0.45.4 download-cosmos-proto: mkdir -p proto/cosmos/base/query/v1beta1 curl https://raw.githubusercontent.com/cosmos/cosmos-sdk/${cosmos-version}/proto/cosmos/base/query/v1beta1/pagination.proto -o proto/cosmos/base/query/v1beta1/pagination.proto mkdir -p proto/google/api curl https://raw.githubusercontent.com/cosmos/cosmos-sdk/${cosmos-version}/third_party/proto/google/api/annotations.proto -o proto/google/api/annotations.proto curl https://raw.githubusercontent.com/cosmos/cosmos-sdk/${cosmos-version}/third_party/proto/google/api/http.proto -o proto/google/api/http.proto mkdir -p proto/gogoproto curl https://raw.githubusercontent.com/cosmos/cosmos-sdk/${cosmos-version}/third_party/proto/gogoproto/gogo.proto -o proto/gogoproto/gogo.proto gen-protoc-ts: download-cosmos-proto install-protoc-gen-ts mkdir -p ./client/src/types/generated/ ls proto/checkers | xargs -I {} protoc \ --plugin="./scripts/node_modules/.bin/protoc-gen-ts_proto" \ --ts_proto_out="./client/src/types/generated" \ --proto_path="./proto" \ --ts_proto_opt="esModuleInterop=true,forceLong=long,useOptionals=messages" \ checkers/{} Makefile View source

Whenever you want to re-compile them, run:

You have created the basic Protobuf objects (opens new window) that will assist you with communicating with the blockchain.

Prepare integration

At this point, you have the generated files in your client folder. If you have made this client folder as a Git submodule, then you can work directly in it and do not need to go back to the checkers Cosmos SDK:

Copy $ cd client

Also, if you use Docker and did not go through the trouble of building the Docker image for the checkers Cosmos SDK, you can use the node:18.7-slim image.

Install the Protobuf.js package in your client project:

At a later stage, you will add checkers as an extension to Stargate, but you can define your checkers extension immediately. The canPlay query could make use of better types for player and position. Start by declaring them in client/src/checkers/player.ts:

Copy export type Player = "b" | "r" export type GamePiece = Player | "*" export interface Pos { x: number y: number } src types checkers player.ts View source

Your checkers extension will need to use the CosmJS Stargate package. Install it:

Now you can declare the checkers extension in src/modules/checkers/queries.ts:

Copy export interface AllStoredGameResponse { storedGames: StoredGame[] pagination?: PageResponse } export interface CheckersExtension { readonly checkers: { readonly getSystemInfo: () => Promise<SystemInfo> readonly getStoredGame: (index: string) => Promise<StoredGame | undefined> readonly getAllStoredGames: ( key: Uint8Array, offset: Long, limit: Long, countTotal: boolean, ) => Promise<AllStoredGameResponse> readonly canPlayMove: ( index: string, player: Player, from: Pos, to: Pos, ) => Promise<QueryCanPlayMoveResponse> } } src modules checkers queries.ts View source

Do not forget a setup function, as this is expected by Stargate:

Copy export function setupCheckersExtension(base: QueryClient): CheckersExtension { const rpc = createProtobufRpcClient(base) // Use this service to get easy typed access to query methods // This cannot be used for proof verification const queryService = new QueryClientImpl(rpc) return { checkers: { getSystemInfo: async (): Promise<SystemInfo> => { const { SystemInfo } = await queryService.SystemInfo({}) assert(SystemInfo) return SystemInfo }, getStoredGame: async (index: string): Promise<StoredGame | undefined> => { const response: QueryGetStoredGameResponse = await queryService.StoredGame({ index: index, }) return response.storedGame }, getAllStoredGames: async ( key: Uint8Array, offset: Long, limit: Long, countTotal: boolean, ): Promise<AllStoredGameResponse> => { const response: QueryAllStoredGameResponse = await queryService.StoredGameAll({ pagination: { key: key, offset: offset, limit: limit, countTotal: countTotal, reverse: false, }, }) return { storedGames: response.storedGame, pagination: response.pagination, } }, canPlayMove: async ( index: string, player: Player, from: Pos, to: Pos, ): Promise<QueryCanPlayMoveResponse> => { return queryService.CanPlayMove({ gameIndex: index, player: player, fromX: Long.fromNumber(from.x), fromY: Long.fromNumber(from.y), toX: Long.fromNumber(to.x), toY: Long.fromNumber(to.y), }) }, }, } } src modules checkers queries.ts View source

You may have to add these imports by hand:

Copy import { assert } from "@cosmjs/utils" import Long from "long" src modules checkers queries.ts View source

Now create your CheckersStargateClient in src/checkers_stargateclient.ts:

Copy export class CheckersStargateClient extends StargateClient { public readonly checkersQueryClient: CheckersExtension | undefined public static async connect( endpoint: string, options?: StargateClientOptions, ): Promise<CheckersStargateClient> { const tmClient = await Tendermint34Client.connect(endpoint) return new CheckersStargateClient(tmClient, options) } protected constructor(tmClient: Tendermint34Client | undefined, options: StargateClientOptions = {}) { super(tmClient, options) if (tmClient) { this.checkersQueryClient = QueryClient.withExtensions(tmClient, setupCheckersExtension) } } } src checkers_stargateclient.ts View source

Integration tests

It is possible to already run some integration tests against a running checkers blockchain.

Preparation

Install packages to run tests.

Describe how to connect to the running blockchain in a .env file in your project root. This depends on where you will run the tests, not on where you run the blockchain:

---

Alternatively, use whichever address connects to the RPC port of the checkers blockchain.

This information will be picked up by the dotenv package. Now let TypeScript know about this in an environment.d.ts file:

Copy declare global { namespace NodeJS { interface ProcessEnv { RPC_URL: string } } } export {} environment.d.ts View source

Also add your tconfig.json as you see fit:

Copy { "exclude": ["./tests/", "./node_modules/", "./dist/"], "compilerOptions": { "esModuleInterop": true, "module": "ES2015", "moduleResolution": "node", "target": "ES6" } } tsconfig.json View source

Add the line that describes how the tests are run:

Copy { ... "scripts": { - "test": "echo \"Error: no test specified\" && exit 1" + "test": "env TS_NODE_COMPILER_OPTIONS='{\"module\": \"commonjs\" }' mocha --require ts-node/register 'test/**/*.ts'" }, ... } package.json View source

First tests

Because the intention is to run these tests against a running chain, possibly a fresh one, they cannot expect too much, such as how many games have been created so far. Still, it is possible to at least test that the connection is made and queries pass through.

Create test/integration/system-info.ts:

Copy import { expect } from "chai" import { config } from "dotenv" import _ from "../../environment" import { CheckersStargateClient } from "../../src/checkers_stargateclient" import { CheckersExtension } from "../../src/modules/checkers/queries" config() describe("SystemInfo", function () { let client: CheckersStargateClient, checkers: CheckersExtension["checkers"] before("create client", async function () { client = await CheckersStargateClient.connect(process.env.RPC_URL) checkers = client.checkersQueryClient!.checkers }) it("can get system info", async function () { const systemInfo = await checkers.getSystemInfo() expect(systemInfo.nextId.toNumber()).to.be.greaterThanOrEqual(1) expect(parseInt(systemInfo.fifoHeadIndex, 10)).to.be.greaterThanOrEqual(-1) expect(parseInt(systemInfo.fifoTailIndex, 10)).to.be.greaterThanOrEqual(-1) }) }) test integration system-info.ts View source

And create one for stored games:

Copy import { expect } from "chai" import { config } from "dotenv" import Long from "long" import _ from "../../environment" import { CheckersStargateClient } from "../../src/checkers_stargateclient" import { CheckersExtension } from "../../src/modules/checkers/queries" config() describe("StoredGame", function () { let client: CheckersStargateClient, checkers: CheckersExtension["checkers"] before("create client", async function () { client = await CheckersStargateClient.connect(process.env.RPC_URL) checkers = client.checkersQueryClient!.checkers }) it("can get game list", async function () { const allGames = await checkers.getAllStoredGames( Uint8Array.of(), Long.fromInt(0), Long.fromInt(0), true, ) expect(allGames.storedGames).to.be.length.greaterThanOrEqual(0) }) it("cannot get non-existent game", async function () { try { await checkers.getStoredGame("no-id") expect.fail("It should have failed") } catch (error) { expect(error.toString()).to.equal( "Error: Query failed with (22): rpc error: code = NotFound desc = not found: key not found", ) } }) }) test integration stored-game.ts View source

Note the forced import of import _ from "../../environment", to actively inform on the string type (as opposed to string | undefined) and avoid any compilation error.

Prepare your checkers chain

There is more than one way to run a checkers blockchain. For instance:

• If you came here after going through the rest of the hands-on exercise, you know how to launch a running chain with Ignite.

• If you arrived here and are only focused on learning CosmJS, it is possible to abstract away niceties of the running chain in a minimal package. For this, you need Docker and to create an image:

  1. Get the Dockerfile:

     Copy $ curl -O https://raw.githubusercontent.com/cosmos/b9-checkers-academy-draft/main/Dockerfile-standalone

  2. Build the image:

     Copy $ docker build . \ -f Dockerfile-standalone \ -t checkersd_i:standalone

If you have another preferred method, make sure to keep track of the required RPC_URL accordingly.

If you are curious about how this Dockerfile-standalone was created, head to the run in production section.

Note that the standalone checkers Docker image uses the project name b9lab/checkers, including for its Protobuf package names. Make sure that your generated Typescript objects do too (opens new window).

If that is not the case, you may encounter ambiguous errors such as:

Copy Query failed with (6): unknown query path: unknown request at QueryClient.queryUnverified (http://localhost:3000/static/js/bundle.js:21568:13) at async Object.getAllStoredGames (http://localhost:3000/static/js/bundle.js:2975:26) at async src_checkers_stargateclient__WEBPACK_IMPORTED_MODULE_0__.CheckersStargateClient.getGuiGames

Launch the tests

Launch your checkers chain. You can choose your preferred method, as long as it can be accessed at the RPC_URL you defined earlier. For the purposes of this exercise, you have the choice between three methods:

---

When using Docker, note:

• --name checkers either matches the name you wrote in RPC_URL, or can be passed as an environment variable to another container to override the value found in .env.
• --network checkers-net, which is reused shortly if you also run your npm tests in Docker. See the paragraph on Docker network, later in this section.

Now, if you run the tests in another shell:

---

This should return:

Copy StoredGame ✔ can get game list (39ms) ✔ cannot get non-existent game SystemInfo ✔ can get system info 3 passing (287ms)

The only combination of running chain / running tests that will not work is if you run Ignite on your local computer and the tests in a container. For this edge case, you should put your host IP address in --env RPC_URL="http://YOUR-HOST-IP:26657".

A note on Docker networks

You may not have used Docker up to this point. The following paragraphs acquaint you with a Docker user-defined bridged network.

If you plan on using Docker Compose at a later stage, having a first taste of such networks is beneficial. Docker Compose can be used to orchestrate and launch separate containers in order to mimic a production setup. In fact, in the production section of this hands-on exercise you do exactly that. If you think this could eventually be useful, you should go through this section. You may want to redo this section with Docker (opens new window).

Earlier you ran the commands:

Copy $ docker network create checkers-net $ docker run --rm -it \ -p 26657:26657 \ --name checkers \ --network checkers-net \ --detach \ checkersd_i:standalone start

This produced the following results:

1. A Docker network was created with the name checkers-net. If containers are started in this network, all ports are mutually accessible.
2. Your container started in it with the resolvable name of checkers.
3. With -p 26657:26657, port 26657 was forwarded to your host computer, on top of being already shared on the checkers-net network.

Then, for tests:

1. When you ran:

   Copy $ npm test

   Your tests, running on the host computer, accessed the checkers chain from the host computer via the forwarded port 26657. Hence RPC_URL="http://localhost:26657".

2. When you ran:

   Copy $ docker run --rm \ -v $(pwd):/client -w /client \ --network checkers-net \ --env RPC_URL="http://checkers:26657" \ node:18.7-slim \ npm test

   Your tests, running in a different container, accessed the checkers chain within the checkers-net Docker network thanks to the checkers name resolution. Hence RPC_URL="http://checkers:26657".

   

   In particular, the -p 26657:26657 port forwarding was not necessary. You can confirm that by stopping your chain and starting it again, this time without -p.

Docker networks are explored further in the next section.

When you are done, if you started the chain in Docker you can stop the containers with:

Copy $ docker stop checkers $ docker network rm checkers-net

synopsis

To summarize, this section has explored:

• The need to prepare the elements that will eventually allow you to create a GUI and/or server-side scripts for your checkers application.
• How to create the necessary Protobuf objects and clients in TypeScript, the extensions that facilitate the use of these clients, so that CosmJS will understand and be able to interact with your checkers module.
• How to use Docker to define a network to launch separate containers that can communicate, for the purpose of integration testing.