refactor: hoist all devDeps to root

[overbalance]

What this does

Hoists all build devDependencies to the root package.json.

• 14 fewer packages in node_modules (1.13% reduction)
• 878 fewer in dependency tree (11.39% reduction)
• 304 fewer packages added by npm ci (10.44% reduction)

Key changes

Test fixes

• Updated karma configs to load plugins from root after dependency hoisting
• Fixed webpack process polyfill path for ESM compatibility
• Updated expect import syntax for socket.io tests
• Updated AWS SDK mock response Content-Length headers

Dependencies

• All build devDependencies moved to root package.json
• Removed unused jQuery dependencies

Configuration

• Converted root .mocharc.yml to .mocharc.json
• Fixed axios import syntax in express example (namespace import → default import)

Package-specific changes

@opentelemetry/instrumentation-socket.io

• Changed import * as expect to import expect (expect v29 uses default export)

@opentelemetry/instrumentation-dns

• Added @ts-expect-error annotations for shimmer __wrapped properties

@opentelemetry/instrumentation-aws-sdk

• Updated Bedrock mock response Content-Length headers

Browser packages (propagator-aws-xray, propagator-instana, instrumentation-user-interaction, instrumentation-long-task)

• Updated karma.conf.js to load plugins from root package.json

[trentm]

Early feedback. I think this is looking pretty good. ;)

• I don't know the karma code well.
• I don't have a feel for the impact of the moduleResolution change in tsconfig.json files.
• (not your fault) Eww on the assert module (because it shadows the core node.js module). Granted that was already installed at the top-level so this PR doesn't impact that at all.

Some questions/notes inline.

[trentm]

Is this one needed?

The openai dep brings in this package, though a MUCH older one.

[email protected] /Users/trentm/src/opentelemetry-js-contrib
└─┬ @opentelemetry/[email protected] -> ./packages/instrumentation-openai
  └─┬ [email protected]
    └── [email protected]

[overbalance]

I previously saw issues when running TAV but I just tried again and it's passing. I'll remove.

[trentm]

Okay, noted. I'll take a look. I know TAV well and I'm the component-owner of instr-openai.

[trentm]

Are we sure we want a change in peer deps here?

[overbalance]

I'm not super familiar with how TAV works but when it installs different versions of redis, it only passes if no single version is specified in dependencies. I would like the owner to review if possible.

[trentm]

Curious why this is the odd package out in needing this.

[overbalance]

In cjs projects, mocha isn't able to process .mjs files without this config.
GCP has this unusual fixture:
argv: ['fixtures/detect-with-http-instrumentation.mjs'],

[trentm]

mocha shouldn't need to process that file, because it is a "fixture": There is a *.test.ts file in this package the exec's that fixture in a subprocess. Similar thing in a number of other packages' test suites.

[overbalance]

Ok. Let me run it without the config and I'll post the error here.

[overbalance]

Working now. Prob another artifact from TS 5.9

[overbalance]

@trentm If Node 18 support is needed then there's more work to do. I missed this requirement when I started this work. 😕 How do you want to proceed? Try to roll back dependency versions until it's happy?

❯ npm ci
npm warn EBADENGINE Unsupported engine {
npm warn EBADENGINE   package: '@azure-rest/[email protected]',
npm warn EBADENGINE   required: { node: '>=20.0.0' },
npm warn EBADENGINE   current: { node: 'v18.20.8', npm: '10.8.2' }
npm warn EBADENGINE }
npm warn EBADENGINE Unsupported engine {
npm warn EBADENGINE   package: '@azure/[email protected]',
npm warn EBADENGINE   required: { node: '>=20.0.0' },
npm warn EBADENGINE   current: { node: 'v18.20.8', npm: '10.8.2' }
npm warn EBADENGINE }

[JamieDanielson]

@trentm If Node 18 support is needed then there's more work to do. I missed this requirement when I started this work. 😕 How do you want to proceed? Try to roll back dependency versions until it's happy?

🤔 We say we support Active and Maintenance LTS versions of Node, and Node 18 is officially end of life. So it might be fine in Contrib to drop support for 18. That's a bigger discussion though (cc @open-telemetry/javascript-maintainers ) and we'll need to update docs too.

[trentm]

If Node 18 support is needed then there's more work to do. I missed this requirement when I started this work. 😕 How do you want to proceed? Try to roll back dependency versions until it's happy?

❯ npm ci
npm warn EBADENGINE Unsupported engine {

@overbalance I don't think those EBADENGINE warnings are blocking for this PR. We already have a number of those warnings in the current state. Unfortunately an EBADENGINE warning will matter for some deps, but not for some deps that are just for testing. For example this one:

npm warn EBADENGINE Unsupported engine {
npm warn EBADENGINE   package: '@azure/[email protected]',
npm warn EBADENGINE   required: { node: '>=20.0.0' },
npm warn EBADENGINE   current: { node: 'v18.20.8', npm: '10.8.2' }
npm warn EBADENGINE }

That dep is used by tedious@17:

% npm ls @azure/core-auth
[email protected] /Users/trentm/src/opentelemetry-js-contrib
└─┬ @opentelemetry/[email protected] -> ./packages/instrumentation-tedious
  └─┬ [email protected]
    ├─┬ @azure/[email protected]
    │ ├── @azure/[email protected]
    │ ├─┬ @azure/[email protected]
    │ │ └── @azure/[email protected] deduped
    │ └─┬ @azure/[email protected]
    │   └── @azure/[email protected] deduped
    └─┬ @azure/[email protected]
      ├─┬ @azure-rest/[email protected]
      │ └── @azure/[email protected] deduped
      ├── @azure/[email protected] deduped
      └─┬ @azure/[email protected]
        └── @azure/[email protected] deduped

which is a test dep. We intentionally install a recent-ish major version of tedious so that unit tests (i.e. npm test) for an instrumentation (instrumentation-tedious in this case) tests with a relatively modern version of the target package. We then rely on our test-all-versions tests to handle testing against those versions of tedious that work with older Node.js versions (e.g. Node.js v18)

[overbalance]

@trentm Those warnings are blocking CI from what I see. What do you recommend?

[trentm]

Those warnings are blocking CI from what I see.

I don't know what is going on. I can npm ci this feature branch locally.

This CI step failed with a network error: https://github.com/open-telemetry/opentelemetry-js-contrib/actions/runs/18537223726/job/53057847496?pr=3032

...
npm WARN EBADENGINE }
npm WARN EBADENGINE Unsupported engine {
npm WARN EBADENGINE   package: '[email protected]',
npm WARN EBADENGINE   required: { node: '20 || >=22' },
npm WARN EBADENGINE   current: { node: 'v18.19.0', npm: '10.2.3' }
npm WARN EBADENGINE }
npm ERR! code ECONNRESET
npm ERR! network aborted
npm ERR! network This is a problem related to network connectivity.
npm ERR! network In most cases you are behind a proxy or have bad network settings.
npm ERR! network 
npm ERR! network If you are behind a proxy, please make sure that the
npm ERR! network 'proxy' config is set properly.  See: 'npm help config'

npm ERR! A complete log of this run can be found in: /home/runner/.npm/_logs/2025-10-17T23_04_44_742Z-debug-0.log
Error: Process completed with exit code 1.

The other test jobs are stuck in npm ci. I cancelled the run after 20m.
I think the "network aborted" above is just a fluke. The real issue is the stuck npm ci steps.

The lint workflow, which does npm ci --ignore-scripts, worked fine.

Ideas on what this could be:

• Some install/build/postinstall script in one of the deps is breaking things.
• Some GitHub Actions or npmjs bug/incident that is causing this to break.
• Some limit is being hit in GH Actions? THis is total guess.