Closed Bug 1959416 Opened 15 days ago Closed 7 days ago

Estimate usage of anchor positioning with different CSS layout models

Categories

(Core :: CSS Parsing and Computation, task)

task

Tracking

()

RESOLVED FIXED

People

(Reporter: jjalkanen, Assigned: jjalkanen)

References

(Blocks 1 open bug)

Details

(Whiteboard: [anchorpositioning:m1])

To better understand the demand for anchor positioning with different layout choices, we can

  • estimate the current use of the feature in grid/flexbox/table/normal layouts
  • find out which of the most often visited online properties rely on anchor positioning and how?
Status: NEW → ASSIGNED
Whiteboard: [anchorpositioning:triage] → [anchorpositioning:m1]

Summary of results based on queries from httparchive:

In average, out of pages on which Blink feature Anchor Positioning is detected

  • With CSSFlexibleBox 2107/2137 ~ 98.6 %
  • With CSSGridLayout 1092/2137 ~ 51 %
  • With CSSGridLayout only 4/2137 ~ 0.19 %
  • Neither 26/2137 ~ 1.2 % (!)

For reference, out of all pages in the same crawl [2],

  • With CSSFlexibleBox 11581430/13185550 ~ 88 %
  • With CSSGridLayout 4201451/13185550 ~ 32 %
  • With CSSGridLayout only 20699/13185550 ~ 0.15 %
  • Neither 1583421/13185550 ~ 12 % (!)

All use also normal block layouts
Table usage is low (or nonexistent)

Note:
Turns out that when a page is exported on a mobile client, all: unset gets expanded into explicit resets for many CSS properties, leading to anchor-name: none and position-anchor: auto definitions. For example, many sites using Feedbackify appear to have this issue, leading to false positives.

Status: ASSIGNED → RESOLVED
Closed: 7 days ago
Resolution: --- → FIXED
You need to log in before you can comment on or make changes to this bug.