Merc6933 GMCI EA - Data Link Feed #3919

Subarna-Singh · 2025-06-25T09:52:06Z

Closes #MERC6933

Description

GMCI EA

to get price value and rebalnce status of GMCI30 Index
uses websocket transportation
EA subscribes to two types of messages
- price [LINK]
- rebalance_status [LINK]
The EA pricesses the incoming messages and stores them in the cache. When data of oth types of messages are available in the cache, the EA combines them to form the EA response.

(https://smartcontract-it.atlassian.net/browse/MERC-6933)

Changes

Create New Adapter for GMCI EA
Override buildConnectionHandlers to allow cacheing price and rebalance_status response, and combine the responses to compose the EA response.

Steps to Test

yarn test packeages/source/gmci

Quality Assurance

If a new adapter was made, or an existing one was modified so that its environment variables have changed, update the relevant infra-k8s configuration file.
If a new adapter was made, or an existing one was modified so that its environment variables have changed, update the relevant adapter-secrets configuration file or update the soak testing blacklist.
[ x] If a new adapter was made, or a new endpoint was added, update the test-payload.json file with relevant requests.
The branch naming follows git flow (feature/x, chore/x, release/x, hotfix/x, fix/x) or is created from Jira.
This is related to a maximum of one Jira story or GitHub issue.
Types are safe (avoid TypeScript/TSLint features like any and disable, instead use more specific types).
All code changes have 100% unit and integration test coverage. If testing is not applicable or too difficult to justify doing, the reasoning should be documented explicitly in the PR.

…js into MERC6933-GMCI-EA-Data_Link_Feed Merge changes from main

changeset-bot · 2025-06-25T09:52:11Z

🦋 Changeset detected

Latest commit: 81abf28

The changes in this PR will be included in the next version bump.

This PR includes changesets to release 1 package

Name	Type
@chainlink/gmci-adapter	Major

Not sure what this means? Click here to learn what changesets are.

Click here if you're a maintainer who wants to add another changeset to this PR

… including field from both messages

mmcallister-cll · 2025-07-09T14:39:00Z

.changeset/itchy-pianos-wave.md

@@ -0,0 +1,5 @@
+---
+'@chainlink/gmci-adapter': minor


Just in case you don't know. You can just edit these changeset files. You don't need to run the command again. Because now you have 2 changeset files which seems unnecessary for a single PR.
Can you remove one of them and make sure the other one is correct?

packages/sources/gmci/package.json

packages/sources/gmci/src/config/overrides.json

packages/sources/gmci/src/endpoint/price.ts

packages/sources/gmci/src/transport/price.ts

packages/sources/gmci/test/integration/fixtures.ts

packages/sources/gmci/test/integration/adapter.test.ts

dskloetc · 2025-07-16T09:55:06Z

packages/sources/gmci/src/transport/price.ts

+              },
+            },
+          }
+          transport.price_cache.set(item.symbol, result, 90000)


Since this is async, there is no guarantee that this result is in the cache by the time you call processMessage.
You should probably just move the entire body here inside processMessage.

dskloetc · 2025-07-16T09:56:37Z

packages/sources/gmci/src/transport/price.ts

+  rebalance_status_cache = new LocalCache<CachedRebalanceResponse>(10)
+}
+
+export async function processMessage(symbols: Array<string>, transport: GMCIWebsocketTransport) {


Can you make this a method on GMCIWebsocketTransport.
It's a bit unfortunate that this function is using transport as a global variable.
If we move it in as a message, it has a cleaner access to everything inside the transport.
Then we still refer to the global transport where the method is called, but at least then it's a single isolated reference.

dskloetc · 2025-07-16T10:01:32Z

packages/sources/gmci/test/unit/price.test.ts

+    expect(message.args).toContain('rebalance_status.gmci30')
+    expect(message.args.length).toBe(2)
+  })
+})


Also add unit tests for the transport itself.

Did you remove it again?

…ing logic to processMessage

dskloetc · 2025-07-16T12:26:42Z

packages/sources/gmci/test/unit/price.test.ts

+      status: 'open',
+    }
+
+    transport.price_cache.set(


Can you mock the websocket connection and get this into the cache by sending the corresponding message over the mock websocket?

dskloetc · 2025-07-16T12:27:48Z

packages/sources/gmci/test/unit/price.test.ts

+      {
+        params: { index: 'GMCI30', type: 'price' },
+        response: {
+          result: 184,


declare variables for the values (price, symbol, timestamps, etc.) so that it's clear that the values you start with are the values you end up with.

dskloetc · 2025-07-16T12:32:01Z

packages/sources/gmci/test/unit/price.test.ts

+
+    expect(mockWrite).toHaveBeenCalledTimes(2)
+    const [callArg] = mockWrite.mock.calls[0]
+    expect(callArg).toBe(transport.name)


Do you know expect(mockWrite).toHaveBeenNthCalledWith(...)?
Then you don't need to reach into the inner workings of the mock.

dskloetc · 2025-07-16T12:36:02Z

packages/sources/gmci/test/unit/price.test.ts

+    expect(result.response.data.status).toBe('open')
+    expect(result.response.data.result).toBe(184)
+    expect(result.response.timestamps.providerIndicatedTimeUnixMs).toBe(1752494400000)
+    expect(result.response.timestamps.providerDataReceivedUnixMs).toBe(now)


I think it's better to expect the value of the entire argument, rather than expect specific individual fields.
That way you know you're not missing something you didn't think of or which was added later.

dskloetc · 2025-07-16T12:37:19Z

packages/sources/gmci/test/unit/price.test.ts

+    expect(result.response.timestamps.providerDataReceivedUnixMs).toBe(now)
+  })
+
+  it('skips writing to cache if any data is missing', async () => {


Also add a test where rebalance status is set but price is not.

dskloetc · 2025-07-16T12:54:29Z

packages/sources/gmci/src/transport/price.ts

+                providerDataStreamEstablishedUnixMs: transport.providerDataStreamEstablished,
+                providerDataReceivedUnixMs: now,
+                providerIndicatedTimeUnixMs:
+                  price_data.response.timestamps.providerIndicatedTimeUnixMs,


What is conceptually the difference between providerDataReceivedUnixMs and providerIndicatedTimeUnixMs?
And why do we use the price_data timestamp as opposed to the rebalance_data timestamp for providerIndicatedTimeUnixMs?

Would it make sense to use the maximum of both timestamps for providerDataReceivedUnixMs, rather than creating a new now timestamp?

dskloetc · 2025-07-16T12:55:37Z

packages/sources/gmci/src/transport/price.ts

+          transport.rebalance_status_cache.get(symbol),
+        ])
+
+        if (cached_price_data != undefined && cached_rebalance_data != undefined) {


If either of them is undefined, return early.
Then you avoid the extra indentation in the rest of the method.

dskloetc · 2025-07-16T13:00:27Z

packages/sources/gmci/src/transport/price.ts

+
+export interface WsResponse {
+  success: boolean
+  data: Array<PriceMessage | RebalanceMessage>


We never mix types of message in a single array, right?
If so data: Array<PriceMessage> | Array<RebalanceMessage> would be more accurate.
Or with the following, the compiler could even deduce the type of array from the value of topic:

interface WsPriceResponse { success: boolean data: Array<PriceMessage> topic: 'price' } interface WsRebalanceResponse { success: boolean data: Array<RebalanceMessage> topic: 'rebalance_status' } export type WsResponse = WsPriceResponse | WsPriceResponse

dskloetc · 2025-07-16T13:02:31Z

packages/sources/gmci/src/transport/price.ts

+}
+
+export interface CachedPriceResponse {
+  params: {


I don't see where these params are used.

dskloetc · 2025-07-16T13:03:04Z

packages/sources/gmci/src/transport/price.ts

+  response: {
+    result: number
+    data: {
+      result: number


This is just an intermediate data object. We don't need to put the same information multiple times in it, right?

…balance_status message - Wintermute does not support rebalance_status

…tractkit/external-adapters-js into MERC6933-GMCI-EA-Data_Link_Feed Merge remote branch changes

Subarna-Singh · 2025-07-17T09:05:41Z

@dskloetc I have a question.
For the WS_API_ENDPOINT, do we keep a default? Since the same EA can now be used for GMCI and Wintermute.

dskloetc · 2025-07-17T09:05:54Z

packages/sources/gmci/src/config/index.ts

+    description: 'WS endpoint for GMCI Data Provider',
+    type: 'string',
+    // default: 'wss://api.gmci.co/private',
+    default: 'wss://generic-alb-ty.api.wintermute-rfq.xyz:15494/private',


Is this for testing?

#3919 (comment)

Thats my question. We should remove the defaultaccording to me. Forces us to set endpoint for GMCI and wintermute separately.

If there is no obvious single default then it makes sense not to have a default at all.

dskloetc · 2025-07-17T09:08:10Z

packages/sources/gmci/src/transport/price.ts

+      if (message.topic === 'price') {
+        for (const item of message.data as PriceMessage[]) {
+          results.push({
+            params: { index: item.symbol },


Why do we call it index if it's the same as symbol? Would it make sense to call it symbol everywhere?

Yup. Makes sense.

dskloetc · 2025-07-17T09:09:43Z

packages/sources/gmci/src/transport/price.ts

@@ -121,44 +160,7 @@ export const options: WebSocketTransportConfig<WsTransportTypes> = {
        return


I don't see a log.

dskloetc · 2025-07-17T09:11:03Z

packages/sources/gmci/src/transport/price.ts

+                symbol: item.symbol,
+              },
+              timestamps: {
+                providerIndicatedTimeUnixMs: convertTimetoUnixMs(item.last_updated),


Weren't there multiple different timestamps before? How do we decide which timestamps to provide?

The others get written here.
Since we had write to the cache ourselves, we have to this bit of work in the previous version.

dskloetc · 2025-07-17T09:13:10Z

packages/sources/gmci/src/transport/price.ts

+      const results = []
+
+      if (message.topic === 'price') {
+        for (const item of message.data as PriceMessage[]) {


I think if we do this, we can avoid a cast here.

dskloetc · 2025-07-17T09:14:18Z

packages/sources/gmci/test/unit/price.test.ts

+    expect(message.args).toContain('rebalance_status.gmci30')
+    expect(message.args.length).toBe(2)
+  })
+})


Did you remove it again?

… Rebalance response

…tractkit/external-adapters-js into MERC6933-GMCI-EA-Data_Link_Feed Merge main branch

dskloetc · 2025-07-18T08:29:13Z

packages/sources/gmci/README.md

+
+| Required? | Name  | Aliases | Description |  Type  | Options | Default | Depends On | Not Valid With |
+| :-------: | :---: | :-----: | :---------: | :----: | :-----: | :-----: | :--------: | :------------: |
+|    ✅     | index |         | Index name  | string |         |         |            |                |


Can you recreate the README now that the name of the parameter is changed?

dskloetc · 2025-07-18T08:34:40Z

packages/sources/gmci/test/unit/price.test.ts

+    expect(results?.[0].params.symbol).toBe('GMCI30')
+    expect(results?.[1].params.symbol).toBe('GML2')
+  })
+})


Can you add a test that rebalance messages don't cause problems?

Subarna-Singh added 4 commits June 19, 2025 19:24

MERC6933-GMCI EA

26760dc

GMCI EA

8d5f2a8

Merge branch 'main' of github.com:smartcontractkit/external-adapters-…

9f4b9f2

…js into MERC6933-GMCI-EA-Data_Link_Feed Merge changes from main

Handle Response

59231d6

Subarna-Singh added 3 commits June 25, 2025 15:23

Changeset added

c06fdd5

Add Readme file; modify tsconfig

3c6f4b5

GMCI EA - parsing price and rebalance messages and producing a result…

60a33d5

… including field from both messages

Subarna-Singh requested a review from mmcallister-cll July 9, 2025 14:22

mmcallister-cll reviewed Jul 9, 2025

View reviewed changes

packages/sources/gmci/package.json Outdated Show resolved Hide resolved

mmcallister-cll reviewed Jul 9, 2025

View reviewed changes

packages/sources/gmci/src/config/overrides.json Outdated Show resolved Hide resolved

mmcallister-cll reviewed Jul 9, 2025

View reviewed changes

packages/sources/gmci/src/endpoint/price.ts Outdated Show resolved Hide resolved

mmcallister-cll reviewed Jul 9, 2025

View reviewed changes

packages/sources/gmci/src/endpoint/price.ts Outdated Show resolved Hide resolved

mmcallister-cll reviewed Jul 9, 2025

View reviewed changes

packages/sources/gmci/src/transport/price.ts Outdated Show resolved Hide resolved

mmcallister-cll reviewed Jul 9, 2025

View reviewed changes

packages/sources/gmci/test/integration/fixtures.ts Show resolved Hide resolved

mmcallister-cll reviewed Jul 9, 2025

View reviewed changes

packages/sources/gmci/test/integration/adapter.test.ts Outdated Show resolved Hide resolved

mmcallister-cll requested a review from a team July 9, 2025 23:15

Subarna-Singh added 10 commits July 11, 2025 16:32

upgrade version of dependency packages

17a621f

remove override functionality

d98103a

Add new changeset

7c0c771

Review work: fix data types and hard coded values

4fe40f2

Integration testing added

000a15e

Remove failure test

ff19d55

Correction to pass: Compile test for changed packages workflow

2e4d6fd

Remove hardcoded values: adapter name

4f55e23

Add gmci to tsconfig files

f4431aa

Merge branch 'main' into MERC6933-GMCI-EA-Data_Link_Feed

e8c8c2b

Subarna-Singh requested a review from mmcallister-cll July 13, 2025 10:24

Testing: add unit tests; add more integration tests

66174dc

Subarna-Singh added 3 commits July 16, 2025 14:48

Remove: old price transport

c727a96

Modify: unit test to reflect new price transport

50d4d42

Fix: uppercase to camelcase

437bfb0

dskloetc reviewed Jul 16, 2025

View reviewed changes

Subarna-Singh added 2 commits July 16, 2025 17:40

Refactor: make processMessage part of wstransport; move message handl…

b9247df

…ing logic to processMessage

Unit test: for price transport

f80198a

Subarna-Singh requested a review from dskloetc July 16, 2025 12:12

Delete .changeset/itchy-pianos-wave.md

0912899

dskloetc reviewed Jul 16, 2025

View reviewed changes

Subarna-Singh added 2 commits July 17, 2025 13:15

Remove: support for rebalance_status message; merging of price and re…

30b8375

…balance_status message - Wintermute does not support rebalance_status

Merge branch 'MERC6933-GMCI-EA-Data_Link_Feed' of github.com:smartcon…

6edb102

…tractkit/external-adapters-js into MERC6933-GMCI-EA-Data_Link_Feed Merge remote branch changes

Subarna-Singh requested a review from dskloetc July 17, 2025 09:04

dskloetc reviewed Jul 17, 2025

View reviewed changes

Subarna-Singh added 10 commits July 17, 2025 18:57

Remove: default API endpoint; Modify: WsResponse to include Price and…

e126611

… Rebalance response

Add unit test for message handler

bf6b59d

Change: index to symbol; make WS_API_ENDPOINT required

418ed2f

Change: tests to reflect symbol instead of index

d4d4720

Change: test payload

871b36d

Merge branch 'main' into MERC6933-GMCI-EA-Data_Link_Feed

dd06ecd

Fix: Failing workflow - missing argument in transport

9916d51

Merge branch 'MERC6933-GMCI-EA-Data_Link_Feed' of github.com:smartcon…

2292dcd

…tractkit/external-adapters-js into MERC6933-GMCI-EA-Data_Link_Feed Merge main branch

Fix mockContext

90ae121

Add: message response type

e28958d

Subarna-Singh requested a review from dskloetc July 18, 2025 08:17

dskloetc reviewed Jul 18, 2025

View reviewed changes

Subarna-Singh added 2 commits July 18, 2025 14:25

Add: unit test for rebalance message

3833941

update readme

81abf28

Subarna-Singh requested a review from dskloetc July 18, 2025 09:41

dskloetc approved these changes Jul 18, 2025

View reviewed changes

		@@ -121,44 +160,7 @@ export const options: WebSocketTransportConfig<WsTransportTypes> = {
		return

Merc6933 GMCI EA - Data Link Feed #3919

Are you sure you want to change the base?

Merc6933 GMCI EA - Data Link Feed #3919

Uh oh!

Conversation

Subarna-Singh commented Jun 25, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Closes #MERC6933

Description

Changes

Steps to Test

Quality Assurance

Uh oh!

changeset-bot bot commented Jun 25, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

🦋 Changeset detected

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Subarna-Singh commented Jul 17, 2025

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Subarna-Singh commented Jun 25, 2025 •

edited

Loading

changeset-bot bot commented Jun 25, 2025 •

edited

Loading