feat(browser otlp exporter): add fetch transport for fetch-only environments

[SacDeNoeuds]

Which problem is this PR solving?

In service workers, XMLHttpRequest and navigator.sendBeacon are both unavailable, making the OTLPTraceExporter from @opentelemetry/exporter-trace-otlp-http unusable.

Fixes #4631

Short description of the changes

Add the fetch transport and enables it only when both xhr and beacons are unavailable.

Type of change

Please delete options that are not relevant.

• Bug fix (non-breaking change which fixes an issue)
• New feature (non-breaking change which adds functionality)
• Breaking change (fix or feature that would cause existing functionality to not work as expected)
• This change requires a documentation update

How Has This Been Tested?

I took the xhr transport tests and adapted the arrange step for the fetch transport.
See the test file: experimental/packages/otlp-exporter-base/test/browser/fetch-transport.test.ts

Checklist:

• Followed the style guidelines of this project
• Unit tests have been added
• Documentation has been updated

Additional context

• feat(otlp-exporter-base): Add fetch sender for ServiceWorker environment #3542 looks a bit old for the project, I gave it another try.

[linux-foundation-easycla]

The committers listed above are authorized under a signed CLA.

• ✅ login: SacDeNoeuds / name: Thomas (fccddd4, ac93c9f, 673af6f, 7994b96, dd4ec25, 63e7a80, 27fe4a9, 23e7330, 4fad1df, daa5b39, 1872256, 35d935e, bc12329, 96bbff6, 008f175, 5530e4e, 0bd231b, dd0b129, c61c687, 84f971e, e3cd53a, 82e013b, cfffa6a, bc4320a, b722a83, d63ab8e, 8a07ae1, 6a3cbfb, f9aeb0a, 14bdf0b, 65da6f9)

[dyladan]

Seems reasonable to me and I don't see any obvious issues. I'd wait for @pichlermarc review before merging because he's much closer to the exporters but overall LGTM

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 95.06%. Comparing base (95b40aa) to head (84f971e).
⚠️ Report is 1 commits behind head on main.

Additional details and impacted files

@@           Coverage Diff           @@
##             main    #5807   +/-   ##
=======================================
  Coverage   95.06%   95.06%           
=======================================
  Files         307      307           
  Lines        8031     8031           
  Branches     1627     1627           
=======================================
  Hits         7635     7635           
  Misses        396      396           

Files with missing lines	Coverage Δ
...-base/src/configuration/otlp-http-configuration.ts	100.00% <100.00%> (ø)

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.
• 📦 JS Bundle Analysis: Save yourself from yourself by tracking and limiting bundle sizes in JS merges.

[SacDeNoeuds]

Hi guys, any news on this?

[raphael-theriault-swi]

We looked at it during the SIG meeting two or three weeks ago but @pichlermarc (who's the de-facto code owner for the exporters) was on vacation at the time. IIRC there was also talks of using fetch for Node.js too / dropping the XHR transport.

[pichlermarc]

IRC there was also talks of using fetch for Node.js too / dropping the XHR transport.

yes, my personal preference would be to drop XHR in favor of fetch as we could increase compatibility with other runtimes while still keeping bundle size roughly the same and keep complexity low.

There has been some push-back to doing this in the past - so I will ask around in the SIG again today and will report back here.

[pichlermarc]

Alright, so looks like we'll move ahead with this PR here and then we'll drop XHR in a later release.

@SacDeNoeuds would you mind marking XhrTransport and associated types as @depreacated? (referring to the newly added FetchTransport as a replacement - with polyfill provided by the end-user if necessary.

I'll create a follow-up issue to tackle removing XHR eventually once this PR has merged.

[SacDeNoeuds]

The former test I wrote was ineffective, every createOtlpXxExporterDelegate returns an instance of the class OTLPExportDelegate, the initial tests were flawed considering they were testing the instance constructor and it was the same constructor – OTLPExportDelegate – in all cases.

I added an intermediary testable function and updated the test.

[pichlermarc]

Thanks for making the changes, just needs some more adjustment on the changelog and we should be good to go. :)

[pichlermarc]

oh - looks like some of the tests in the individual exporters now need to be adapted too as they expect XHR to be used. There is no need to test all cases in the individual exporters as we do test that behavior in the base package already, but we should change the tests from XHR to fetch there.

[SacDeNoeuds]

Thanks again for your reviews, I took a bit of time to update all the tests, all tests should pass now.

It may be worth adding the script npm run test:browser along with npm run test in the CONTRIBUTING guide, I assumed npm run test was covering both server and browser environments.

[pichlermarc]

Looks good, thank you for working on this 🙌

[SacDeNoeuds]

Thanks @pichlermarc, I just fixed a small changelog conflict. I suppose you will be dealing with PR merging?

[opentelemetrybot]

Thank you for your contribution @SacDeNoeuds! 🎉 We would like to hear from you about your experience contributing to OpenTelemetry by taking a few minutes to fill out this survey.