💥 Remove FID (First Input Delay) tracking

[allspain]

Motivation

FID (First Input Delay) was deprecated as a Core Web Vital in March 2024, replaced by INP (Interaction to Next Paint). Since v7 allows breaking changes, this is the right time to remove FID tracking, its polyfill, and all related event fields.

Changes

• Deleted: trackFirstInput.ts, trackFirstInput.spec.ts, firstInputPolyfill.ts
• Schema: Removed fid from _view-performance-schema.json, removed first_input_delay, first_input_time, first_input_target_selector from view-schema.json
• Samples: Removed FID fields from view.json and stream.json
• Types: Regenerated rumEvent.types.ts; removed FID fields from rawRumEvent.types.ts and ViewPerformanceData
• performanceObservable.ts: Removed polyfill import and fallback block. Preserved FIRST_INPUT enum, RumFirstInputTiming interface, and EntryTypeToReturnType mapping (used by INP)
• trackInitialViewMetrics.ts: Removed trackFirstInput call and firstInput from metrics
• viewCollection.ts: Removed FID field assignments and fid from performance data
• E2E: Removed "send performance first input delay" test
• BROWSER_SUPPORT.md: Updated web vitals row (Firefox/Safari no longer have FID-only support), removed FID footnote

Test instructions

• yarn typecheck — passes
• yarn lint — passes (1 pre-existing warning)
• yarn format — passes
• yarn test:unit — 3199/3200 pass (1 pre-existing skip)
• INP tests (trackInteractionToNextPaint.spec.ts) — all 18 pass, confirming FIRST_INPUT entries still work

Checklist

• Tested locally
• Tested on staging
• Added unit tests for this change.
• Added e2e/integration tests for this change.
• Updated documentation and/or relevant AGENTS.md file

🤖 Generated with Claude Code

[github-actions]

Thank you for your submission, we really appreciate it. Like many open-source projects, we ask that you sign our Contributor License Agreement before we can accept your contribution. You can sign the CLA by just posting a Pull Request Comment same as the below format.

---

I have read the CLA Document and I hereby sign the CLA

---

_{You can retrigger this bot by commenting recheck in this Pull Request. Posted by the CLA Assistant Lite bot.}

[cit-pr-commenter-54b7da]

Bundles Sizes Evolution

📦 Bundle Name	Base Size	Local Size	𝚫	𝚫%	Status
Rum	170.35 KiB	168.91 KiB	-1.45 KiB	-0.85%	✅
Rum Profiler	4.69 KiB	4.69 KiB	0 B	0.00%	✅
Rum Recorder	24.91 KiB	24.91 KiB	0 B	0.00%	✅
Logs	56.38 KiB	56.38 KiB	0 B	0.00%	✅
Flagging	944 B	944 B	0 B	0.00%	✅
Rum Slim	127.63 KiB	126.25 KiB	-1.38 KiB	-1.08%	✅
Worker	23.63 KiB	23.63 KiB	0 B	0.00%	✅

🚀 CPU Performance

Action Name	Base CPU Time (ms)	Local CPU Time (ms)	𝚫%
RUM - add global context	0.0063	0.0066	+4.76%
RUM - add action	0.0136	0.0187	+37.50%
RUM - add error	0.0142	0.0168	+18.31%
RUM - add timing	0.0029	0.0039	+34.48%
RUM - start view	0.0124	0.0206	+66.13%
RUM - start/stop session replay recording	0.0007	0.0013	+85.71%
Logs - log message	0.0146	0.0232	+58.90%

🧠 Memory Performance

Action Name	Base Memory Consumption	Local Memory Consumption	𝚫
RUM - add global context	26.93 KiB	27.13 KiB	+204 B
RUM - add action	115.44 KiB	111.16 KiB	-4.27 KiB
RUM - add timing	25.89 KiB	26.09 KiB	+211 B
RUM - add error	120.19 KiB	117.99 KiB	-2.20 KiB
RUM - start/stop session replay recording	25.57 KiB	26.33 KiB	+786 B
RUM - start view	506.11 KiB	507.74 KiB	+1.63 KiB
Logs - log message	45.61 KiB	46.60 KiB	+1009 B

🔗 RealWorld

[datadog-datadog-prod-us1]

✅ Tests

🎉 All green!

❄️ No new flaky tests detected
🧪 All tests passed

🎯 Code Coverage (details)
• Patch Coverage: 100.00%
• Overall Coverage: 77.16% (+0.19%)

_{This comment will be updated automatically if new data arrives.
🔗 Commit SHA: 364a494 | Docs | Datadog PR Page | Was this helpful? Give us feedback!}

[allspain]

RFC: Remove FID (First Input Delay) Tracking

Status: Complete

Branch: richard.klein/remove-fid

Last updated: 2026-02-26

PR: PR #4223

Summary

This change removes First Input Delay (FID) tracking from the Browser SDK. FID was deprecated as a Core Web Vital by Google in March 2024, replaced by Interaction to Next Paint (INP). Since the v7 branch allows breaking changes, this is the appropriate time to remove the FID implementation, its polyfill for older browsers, and all code that populates FID-related event fields.

The event format schema is preserved for backwards compatibility — FID fields remain defined in the schema but the SDK no longer populates them.

Motivation

FID measured the delay between a user's first interaction and the browser's response. Google deprecated FID as a Core Web Vital in March 2024, replacing it with INP which measures responsiveness across all interactions, not just the first one.

The SDK already tracks INP via trackInteractionToNextPaint.ts. Continuing to track FID adds dead code, an unnecessary polyfill (firstInputPolyfill.ts) for browsers that don't support the first-input performance entry natively, and event fields that are no longer meaningful.

Important constraint: INP tracking (trackInteractionToNextPaint.ts) subscribes to FIRST_INPUT performance entries and uses the RumFirstInputTiming interface. The FIRST_INPUT enum value, RumFirstInputTiming interface, and related type mappings in performanceObservable.ts must be preserved.

Solution

Approach

Remove the FID tracking implementation and polyfill while preserving the shared performance entry infrastructure used by INP. The event format schema retains FID fields for backwards compatibility — the SDK simply stops populating them.

Architecture

The FID tracking pipeline was:

1. performanceObservable.ts — observed first-input performance entries (with polyfill fallback)
2. trackFirstInput.ts — subscribed to the observable, computed delay, extracted target selector
3. trackInitialViewMetrics.ts — wired trackFirstInput into the initial view metrics collection
4. viewCollection.ts — serialized firstInput metrics into first_input_delay, first_input_time, first_input_target_selector, and performance.fid

This change removes steps 2-4 entirely, and removes the polyfill fallback from step 1. The FIRST_INPUT performance entry observation itself remains because INP uses it.

Schema

The RumFirstInputTiming interface in performanceObservable.ts (preserved for INP):

export interface RumFirstInputTiming {
  entryType: RumPerformanceEntryType.FIRST_INPUT
  startTime: RelativeTime
  processingStart: RelativeTime
  processingEnd: RelativeTime
  duration: Duration
  target?: Node
  interactionId?: number
  toJSON(): Omit<RumFirstInputTiming, 'toJSON'>
}

The FIRST_INPUT enum value and EntryTypeToReturnType mapping are also preserved.

Types removed from RawRumViewEvent.view in rawRumEvent.types.ts:

first_input_delay?: ServerDuration
first_input_time?: ServerDuration
first_input_target_selector?: string

Types removed from ViewPerformanceData in rawRumEvent.types.ts:

fid?: {
  duration: ServerDuration
  timestamp: ServerDuration
  target_selector?: string
}

Testing

• yarn typecheck — passes
• yarn lint — passes (1 pre-existing warning)
• yarn format — passes
• yarn test:unit — 3199/3200 pass (1 pre-existing skip)
• INP tests (trackInteractionToNextPaint.spec.ts) — all 18 pass, confirming FIRST_INPUT performance entries still work correctly
• viewCollection.spec.ts — 9/9 pass with FID fields removed from fixtures and assertions
• performanceObservable.spec.ts — 7/7 pass

References

• PR #4223
• Google: Retiring FID — March 2024 announcement
• web.dev FID documentation — original FID specification
• web-vitals reference implementation — reference that trackFirstInput.ts was based on

[BeltranBulbarellaDD]

LGTM!
Some format things we can discuss:

1. In Changes part, I will see the changes in the modified files, I think I would prefer to see a general idea of the changes in a paragraph more than bullet points of each change.
2. Same for Test instructions. This part is for me to know how to test it. That the CI passes I can see it below. Maybe in this case that is removing code you can say: typecheck passes. Since it's the relevant test.
3. The RFC part sort of repeats what is in the description. I would prefer it all to be in the description since it's the first thing someone reads.
4. I like the RFC, I think it's a little bit long. Specially around the Architecture part.

Overall IMO, it's nice to read all this info. I completely understand what you try to do and I have sufficient context to review!