Implement composite cache by mxiao-cll · Pull Request #765 · smartcontractkit/ea-framework-js

mxiao-cll · 2026-05-07T23:34:31Z

Background: https://smartcontract-it.atlassian.net/browse/OPDATA-6402

Commit 1:

Split ResponseCache into ResponseCache (base) and SimpleResponseCache (implementation)
Move constructor into base
Move writeTTL into base
Move cache entry creation logic into base
Everything else is in implementation

Commit 2:

Re-structure cache so that we can use different transport name for cache key & meta in EA response field
- If we have a composite transport holding http and ws transport, we still want http and ws transport to write into the same cache key, however the EA response should reflect which transport is used to generate this answer
Implement compare cache
- It holds a responseCache which actually writes into cache
- It also takes a shouldUpdate function to determine if new entry should be written into cache or not
- Instead of always loading from cache, it maintains a list of recently written entry so we do a local comparison first
- It uses responseCache's transport for cache key but it's own transport for EA response meta
Implement composite transport
- Initialize all the child transport with the same compare cache
- Simply forward each operation to the individual child transport and let cache do the work

Commit 3:

Add docs

How to use this: https://github.com/smartcontractkit/external-adapters-js/pull/4939/changes

github-actions · 2026-05-07T23:34:43Z

👋 mxiao-cll, thanks for creating this pull request!

To help reviewers, please consider creating future PRs as drafts first. This allows you to self-review and make any final changes before notifying the team.

Once you're ready, you can mark it as "Ready for review" to request feedback. Thanks!

github-actions · 2026-05-07T23:34:53Z

NPM Publishing labels 🏷️

🟢 This PR has valid version labels and will cause a patch bump.

alejoberardino

@mxiao-cll from the ticket for this change I agree with your challenge arguments wholeheartedly, would be good to see the intended effect of this change in the concrete impl you have in mind even as a draft

mxiao-cll · 2026-05-08T03:32:17Z

@mxiao-cll from the ticket for this change I agree with your challenge arguments wholeheartedly, would be good to see the intended effect of this change in the concrete impl you have in mind even as a draft

Big PR here you go and example usage here: https://github.com/smartcontractkit/external-adapters-js/pull/4939/changes

alejoberardino · 2026-05-12T03:09:41Z

+  async writeEntries() {
+    throw new Error('Use write instead for CompareResponseCache')
+  }


this one seems to be internal, can probably just be a method of the simple class and not of this one instead of being part of the interface

The problem is that the compareCache takes the interface as input instead of the simple class

seems more correct to me to have this expect an extended interface (or even the class) than to have this fail at runtime

alejoberardino · 2026-05-12T03:16:58Z

+})
+
+test.serial(
+  'composite transport merges child writes using shouldUpdate when run under an adapter',


would be nice to add a couple more test cases for the unhappy paths, e.g. having one transport fail to produce a value, requests coming with a specific transport request, etc

Added 2 tests, not sure about "requests coming with a specific transport request", that doesn't seem to be related to the new transport added.

Unless we took it out the framework supports specifying which transport to use when requesting (useful when the EA for example has a crypto endpoint and we know in advance a particular asset pair only exists in the rest transport and not the ws). I don't think this is a feature we can / should remove, and for safety I'd probably have the framework make sure that cases like these are supported, working, and not allowing for multiple instances of the transports that could cause problems (e.g. multiple ws connections, rate limiting failures to the provider)

alejoberardino · 2026-05-13T19:47:27Z

+  async writeEntries() {
+    throw new Error('Use write instead for CompareResponseCache')
+  }


seems more correct to me to have this expect an extended interface (or even the class) than to have this fail at runtime

alejoberardino · 2026-05-13T19:50:08Z

+    await Promise.all(
+      Object.entries(this.config.transports).map(([name, transport]) =>
+        transport.initialize(
+          { ...dependencies, responseCache: compareCache },
+          adapterSettings,
+          endpointName,
+          name,
+        ),
+      ),
+    )


One thing that came to mind when looking at the example usage PR in the ea monorepo (thanks for that btw, made this easier to reason even if very simple) is that we could mistakenly create multiple instances of the same adapter and initialize both, and potentially do unexpected things from the EAs like creating multiple connections to providers when we only intended/support one.

We need to add something on the framework side to check that if we're using a composite transport for a particular endpoint, either the underlying transports are the same as the composite one, we take the composite one and register the underlying ones automatically, etc.

alejoberardino · 2026-05-13T19:53:07Z

+})
+
+test.serial(
+  'composite transport merges child writes using shouldUpdate when run under an adapter',


Unless we took it out the framework supports specifying which transport to use when requesting (useful when the EA for example has a crypto endpoint and we know in advance a particular asset pair only exists in the rest transport and not the ws). I don't think this is a feature we can / should remove, and for safety I'd probably have the framework make sure that cases like these are supported, working, and not allowing for multiple instances of the transports that could cause problems (e.g. multiple ws connections, rate limiting failures to the provider)

mxiao-cll requested a review from a team as a code owner May 7, 2026 23:34

mxiao-cll added the none label May 7, 2026

mxiao-cll removed request for LunkRat, fmunshi, johnnymugs, mohamed-mehany and vlfig May 7, 2026 23:40

mxiao-cll requested review from alejoberardino and removed request for chray-zhang, chris-chainlink, cl-efornaciari, cll-gg, danwilliams-cll, dskloetc, eduard-cl, florian-cl, karen-stepanyan, mmcallister-cll and thomjg May 7, 2026 23:40

alejoberardino reviewed May 8, 2026

View reviewed changes

mxiao-cll requested a review from alejoberardino May 8, 2026 03:32

mxiao-cll force-pushed the OPDATA-6402 branch 2 times, most recently from ca83238 to a414c9e Compare May 8, 2026 13:12

mxiao-cll changed the title ~~Create a base class for response cache~~ Implement composite cache May 8, 2026

mxiao-cll force-pushed the OPDATA-6402 branch from e60af5e to 5ce639b Compare May 8, 2026 13:32

mxiao-cll added patch and removed none labels May 8, 2026

mxiao-cll mentioned this pull request May 8, 2026

Temp smartcontractkit/external-adapters-js#4939

Draft

alejoberardino reviewed May 12, 2026

View reviewed changes

mxiao-cll added 4 commits May 12, 2026 09:04

Create a base class for response cache

3590af2

compare cache & composite transport

158f119

docs

6f6f1c3

Comments

42e0893

mxiao-cll force-pushed the OPDATA-6402 branch from 5ce639b to 42e0893 Compare May 12, 2026 13:49

mxiao-cll requested a review from alejoberardino May 12, 2026 13:51

alejoberardino requested changes May 13, 2026

View reviewed changes

Conversation

mxiao-cll commented May 7, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

github-actions Bot commented May 7, 2026

Uh oh!

github-actions Bot commented May 7, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

NPM Publishing labels 🏷️

Uh oh!

alejoberardino left a comment

Choose a reason for hiding this comment

Uh oh!

mxiao-cll commented May 8, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

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!

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!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

mxiao-cll commented May 7, 2026 •

edited

Loading

github-actions Bot commented May 7, 2026 •

edited

Loading

mxiao-cll commented May 8, 2026 •

edited

Loading