Create presentation-timestamps explainer #106
Conversation
There was a problem hiding this comment.
This document seems to imply that Chromium should start tracking the timing of the end of "update the rendering". Do you also expect that Chromium should change the times reported in existing attributes to match that, or should we update those specs to declare that the renderTime / startTime attributes refer to these new timestamps instead?
presentation-timestamps.md
Outdated
| When originally introduced, paint timing ("`first-paint`" and "`first-contentful-paint`") aspired to represent an important moment in terms of user experience - "The pixels on the screen" representing a certain state. | ||
| However, this is tricky in terms of interopability - that moment is not always measured in the same way, and is not part of the flow of operations covered by web standards. | ||
| Originally in the paint-timing spec, a different time is used instead as the `startTime` - the end of the "update the rendering" phase, where the document is done setting up the rendering and hands over rendering to the user agent. | ||
| This is confusing and non-interoperable, as Chromium still reports the VSync-time, which sufficiently to corresponds to "pixels on the screen". |
There was a problem hiding this comment.
There's an extra "to" here -- I think either "sufficiently corresponds to" or "corresponds sufficiently to" is intended
presentation-timestamps.md
Outdated
| ## Overview | ||
| When originally introduced, paint timing ("`first-paint`" and "`first-contentful-paint`") aspired to represent an important moment in terms of user experience - "The pixels on the screen" representing a certain state. | ||
| However, this is tricky in terms of interopability - that moment is not always measured in the same way, and is not part of the flow of operations covered by web standards. | ||
| Originally in the paint-timing spec, a different time is used instead as the `startTime` - the end of the "update the rendering" phase, where the document is done setting up the rendering and hands over rendering to the user agent. |
There was a problem hiding this comment.
This is still the case - we call "mark paint timing" from HTML, which sets a time for FP, FCP, LCP and ET. Is only LoAF which is different?
Also, can this link to the HTML spec?
presentation-timestamps.md
Outdated
| - `first-contentful-paint` | ||
| - `long-animation-frame` | ||
|
|
||
| It would the following shape: |
There was a problem hiding this comment.
"It would have the following shape:"
presentation-timestamps.md
Outdated
|
|
||
| ## Conclusion | ||
|
|
||
| When exposing paint timings, we look for the right trade-off between "UX-percise" and "interoperable". |
Oops I'll add that to the explainer. The existing timings return |
It's actually already the result of the discussion on the WebPerfWG (#62) |
SHA: 52d10fc Reason: push, by noamr Co-authored-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com>
Creating an explainer to help with TAG and other reviews.