✨ feat: collect LCP subparts RUM-13818

[HugoGarrido]

Motivation

LCP Subparts are values that compute the final LCP measure the browser SDK.

This is helpful for troubleshooting LCP since it help to define what type of LCP issue we are facing (server, render etc.) by looking at the biggest contributor (more details in this doc)

Here is a breakdown of each part :

Subpart	Meaning
TTFB	Already tracked by the Browser SDK as firstByte
Resource load delay	Time between firstByte and the loading start of the resource associated with the LCP. 0 if no resource are attached to the LCP
Resource load duration	Time to takes to load the resource attached to the LCP0 if no resource are attached to the LCP
Element render delay	Time between the LCP resource finishes loading and the LCP element is fully rendered

This PR adds the collection of the LCP subparts when we collect the LCP

Changes

• In trackLargestContentfulPaint when we have an LCP entry :
  • re-measure the TTFB (keeping the name firstByte)
  • get the performance entry related to the LCP resource if applicable
  • compute an optional subParts object that is added in the final payload
• Extract a helper to safely compute the firstByte value
• Extract a helper to safely read performance global
• Update tests, fixtures and types to support the new subParts object

Test instructions

Checklist

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

[github-actions]

All contributors have signed the CLA ✍️ ✅
_{Posted by the CLA Assistant Lite bot.}

[cit-pr-commenter-54b7da]

Bundles Sizes Evolution

📦 Bundle Name	Base Size	Local Size	𝚫	𝚫%	Status
Rum	164.36 KiB	164.97 KiB	+621 B	+0.37%	✅
Rum Profiler	4.33 KiB	4.33 KiB	0 B	0.00%	✅
Rum Recorder	24.27 KiB	24.27 KiB	0 B	0.00%	✅
Logs	56.14 KiB	56.14 KiB	0 B	0.00%	✅
Flagging	944 B	944 B	0 B	0.00%	✅
Rum Slim	121.57 KiB	122.14 KiB	+592 B	+0.48%	✅
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	0.0042	0.0041	-2.38%
RUM - add action	0.0135	0.0132	-2.22%
RUM - add error	0.0131	0.0126	-3.82%
RUM - add timing	0.0029	0.0028	-3.45%
RUM - start view	0.0043	0.0033	-23.26%
RUM - start/stop session replay recording	0.0008	0.0007	-12.50%
Logs - log message	0.0147	0.0143	-2.72%

🧠 Memory Performance

Action Name	Base Memory Consumption	Local Memory Consumption	𝚫
RUM - add global context	26.27 KiB	25.69 KiB	-591 B
RUM - add action	48.61 KiB	48.66 KiB	+48 B
RUM - add timing	24.97 KiB	26.06 KiB	+1.09 KiB
RUM - add error	54.76 KiB	55.11 KiB	+361 B
RUM - start/stop session replay recording	23.95 KiB	25.19 KiB	+1.25 KiB
RUM - start view	424.85 KiB	426.04 KiB	+1.19 KiB
Logs - log message	95.72 KiB	44.24 KiB	-51.48 KiB

🔗 RealWorld

[datadog-official]

✅ Tests

🎉 All green!

❄️ No new flaky tests detected
🧪 All tests passed

🎯 Code Coverage
• Patch Coverage: 69.23%
• Overall Coverage: 77.29% (+0.03%)

View detailed report

_{This comment will be updated automatically if new data arrives.
🔗 Commit SHA: e0dd7ee | Docs | Datadog PR Page | Was this helpful? Give us feedback!}

[HugoGarrido]

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

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 94434b0253

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

[chatgpt-codex-connector]

Match the LCP resource to the correct request

The LCP resource timing is selected via getResourceEntries()?.find((e) => e.name === resourceUrl), which returns the first entry with that URL. Resource entries persist for the whole page and can include multiple requests for the same URL (e.g., preloads, cache-busting reloads, or SPA route changes). In those cases the first entry is often unrelated to the LCP element, so loadDelay/loadTime/renderDelay are computed from stale timings, corrupting the subparts metrics. Consider picking the most recent matching entry whose timing precedes lcpEntry.startTime instead of the first match.

Useful? React with 👍 / 👎.

[chatgpt-codex-connector]

Avoid reporting subparts when firstByte is invalid

When getSafeFirstByte() returns undefined for negative or future responseStart values (the exact cases it is meant to guard against), the code falls back to 0 and still emits subparts. This will silently report a 0ms TTFB and inflate renderDelay for affected browsers, which is misleading data rather than a safe omission. Consider skipping subParts or leaving firstByte undefined when getSafeFirstByte() fails.

Useful? React with 👍 / 👎.