Tweak LCP render time wording

[tunetheweb]

Description

I noticed some inaccuracies in the LCP render time page:

• The top of the page states 0 will be returned for cross-origin requests but this is not the case now.
• Therefore you cannot use a non-zero render time as a measure whether the LCP is accurate.
• Reword the startTime section as I found that difficult to understand.

I changed this in #37743 but missed these.

Motivation

Correct inaccuracies to avoid confusion

Additional details

Related issues and pull requests

Relates to: #37743
Relates to: w3c/largest-contentful-paint#138

[github-actions]

Preview URLs

• /en-US/docs/Web/API/LargestContentfulPaint/renderTime
• /en-US/docs/Web/API/LargestContentfulPaint

External URLs (1)

URL: /en-US/docs/Web/API/LargestContentfulPaint
Title: LargestContentfulPaint

• https://github.com/w3c/paint-timing/issues/104 (1 time) (Note! This may be a new URL 👀)

(comment last updated: 2025-09-13 10:35:52)

[jasonwilliams]

is https://developer.mozilla.org/en-US/docs/Web/API/PerformanceEntry/startTime#largest-contentful-paint still accurate?

[tunetheweb]

is https://developer.mozilla.org/en-US/docs/Web/API/PerformanceEntry/startTime#largest-contentful-paint still accurate?

Yes it is. In most cases now the renderTime will no longer be zero, but old Chromes will be. Plus we don't know yet how Safari will handle this (or any other browser engines).

Ultimately the wording there is still specified and also still matches the changes proposed here which still mention it might be zero.

[wbamberg]

I don't understand why the "### Cross-origin image render time" is duplicated. Is there a reason to not just have it in the page for renderTime?

[tunetheweb]

I don't understand why the "### Cross-origin image render time" is duplicated. Is there a reason to not just have it in the page for renderTime?

This is a somewhat fair comment. However, I'd prefer to have it duplicated for a couple of reasons:

1. We're talking about what to use for LCP time. So although this nuance is in regards to renderTime, it affects the over all value to be used for this API.
2. We recommend kinda ignoring renderTime anyway. So to bury this information in that page feels kinda like... well burying it.

Does that convince you at all that duplication, while not ideal, is important enough to have this on the top page?

[wbamberg]

I don't understand why the "### Cross-origin image render time" is duplicated. Is there a reason to not just have it in the page for renderTime?

This is a somewhat fair comment. However, I'd prefer to have it duplicated for a couple of reasons:

1. We're talking about what to use for LCP time. So although this nuance is in regards to renderTime, it affects the over all value to be used for this API.

2. We recommend kinda ignoring renderTime anyway. So to bury this information in that page feels kinda like... well burying it.

Does that convince you at all that duplication, while not ideal, is important enough to have this on the top page?

This is a good point. But it's quite a complex bit of exposition and I do think duplicating it is tough on readers. As a reader I have to try to work out if this is actually exactly the same or if there are subtle differences I need to worry about. And then perhaps some maintainer will change one and not the other, and now there is a difference, and readers have to decide whether the difference is intentional or not.

My suggestion is to add a note to the interface page somewhat like:

To measure render time for cross-origin requests, set the {{httpheader("Timing-Allow-Origin")}} header.

To ensure more accurate results for cross-origin requests across all browsers, developers should use startTime instead of renderTime as the LCP value.

See Cross-origin image render time and Use startTime over renderTime for more details.

But I won't block approval on this, if you disagree.

[wbamberg]

There's a sentence fragment here. I didn't know if you also wanted to keep the preceding sentence on "Instead..." - I have removed it in my suggestion because it seems to me to be covered in the next section ("Use startTime") and because I think "Browsers may now" flows better without it. But your call.

Suggested change

	For security reasons, the value of the {{domxref("LargestContentfulPaint.renderTime", "renderTime")}} property was originally `0` if the resource is a cross-origin request. Instead the {{domxref("LargestContentfulPaint.loadTime", "loadTime")}} property should be used as a fallback. The {{domxref("PerformanceEntry.startTime", "startTime")}}
	For security reasons, the value of the {{domxref("LargestContentfulPaint.renderTime", "renderTime")}} property was originally `0` if the resource is a cross-origin request.

[tunetheweb]

OK I've removed this whole section from the LCP page and referenced the section in the renderTime page as suggested.

So I made this suggested change in that page instead.

[wbamberg]

This link goes to the interface BCD table, which doesn't show detailed compat for renderTime.

[tunetheweb]

Yes it does under cross-origin under renderTime:

I'm not convinced this is obvious to developer that this is what this represents but was asked to change it from my original suggestion of cross-origin-render-time

[tunetheweb]

OK reworked it to have a lighter explanation on the LCP page and link to the heavier explanation on the renderTime page. PTAL!