👷 change e2e implementation for workers

[mormubis]

Motivation

The initial implementation of Worker E2E tests was somewhat tied to our specific use case and will not scale if we need to test different scenarios within the Worker environment.

Changes

It introduces a version of withWorker that automatically sets up the logs inside the worker. We can also pass an implementation, similar to what we do with withLogs and withBody.

Test instructions

All e2e still passed.

Checklist

• Tested locally
• Tested on staging
• Added unit tests for this change.
• Added e2e/integration tests for this change.

[cit-pr-commenter]

Bundles Sizes Evolution

📦 Bundle Name	Base Size	Local Size	𝚫	𝚫%	Status
Rum	162.08 KiB	162.08 KiB	0 B	0.00%	✅
Rum Recorder	19.78 KiB	19.78 KiB	0 B	0.00%	✅
Rum Profiler	4.89 KiB	4.89 KiB	0 B	0.00%	✅
Logs	55.77 KiB	55.77 KiB	0 B	0.00%	✅
Flagging	944 B	944 B	0 B	0.00%	✅
Rum Slim	119.06 KiB	119.06 KiB	0 B	0.00%	✅
Worker	23.60 KiB	23.60 KiB	0 B	0.00%	✅

🚀 CPU Performance

Action Name	Base CPU Time (ms)	Local CPU Time (ms)	𝚫%
RUM - add global context	0.0045	0.0074	+64.44%
RUM - add action	0.014	0.0147	+5.00%
RUM - add error	0.0122	0.0126	+3.28%
RUM - add timing	0.003	0.0031	+3.33%
RUM - start view	0.0037	0.004	+8.11%
RUM - start/stop session replay recording	0.001	0.0008	-20.00%
Logs - log message	0.0176	0.0133	-24.43%

🧠 Memory Performance

Action Name	Base Memory Consumption	Local Memory Consumption	𝚫
RUM - add global context	25.46 KiB	25.93 KiB	+477 B
RUM - add action	45.97 KiB	45.12 KiB	-872 B
RUM - add timing	24.34 KiB	24.92 KiB	+596 B
RUM - add error	51.44 KiB	49.84 KiB	-1.60 KiB
RUM - start/stop session replay recording	24.89 KiB	24.30 KiB	-608 B
RUM - start view	422.83 KiB	426.77 KiB	+3.94 KiB
Logs - log message	43.14 KiB	42.29 KiB	-870 B

🔗 RealWorld

[datadog-official]

✅ Tests

🎉 All green!

❄️ No new flaky tests detected
🧪 All tests passed

🎯 Code Coverage
• Patch Coverage: 100.00%
• Total Coverage: 92.67% (+0.00%)

View detailed report

_{This comment will be updated automatically if new data arrives.
🔗 Commit SHA: 9afa674 | Docs | Was this helpful? Give us feedback!}

[BeltranBulbarellaDD]

Would it be possible to move this to the createTest? So that we do something like:
.withWorker({ isModule: true, handleMessageWith: 'logger' })
Similar to what we do in:

browser-sdk/test/e2e/scenario/browser-extensions/browserExtensions.scenario.ts

Line 22 in 35fd80d

.withExtension(createExtension(path.join(BASE_PATH, name)).withRum().withLogs())

[mormubis]

I think the idea is to avoid that scenario and be more generic, or that's what I understood from @bcaudan feedback.

[bcaudan]

I think we could make the API more convenient by:

• providing a function to withWorker, to work directly with code instead of a block of text
• have the worker implementation as complete as possible:
  • let it import the script rather than handle that in workerSetup, it would be easier to follow
  • provide it the consolidated configuration

to be able to do something like:

.withWorker((config) => {
  import '/datadog-logs.js'
      
  // Initialize DD_LOGS in service worker
  DD_LOGS.init(config)

  // Handle messages from main thread
  self.addEventListener('message', (event) => {
    const message = event.data
    DD_LOGS.logger.log(message)
  })
})

[mormubis]

@bcaudan I had this as an earlier version, but then I found that the current implementation is closer to what we do with Logs/RUM (withLogs and withRUM).

[bcaudan]

❓ question: ‏Do we need to use post message for every test? Could we instead have the log needed for the test be triggered when the worker is started from the worker implementation?
It would reduce the complexity of those tests

[mormubis]

@bcaudan I thought about it, and it's a bit tricky, because it means the test it auto-executes it. Are you ok with that?

[bcaudan]

From my understanding the test would instantiate the worker and this instantiation could generate the log use case that we want to test, avoiding the extra message exchanges with the main thread.
Was there any issue that this message exchange was preventing or any other issue that you see with the "auto-execution"?
if not, I'd say it is worth a try.

[mormubis]

Done!

[bcaudan]

Could we do it with other tests as well?