`-webkit-fill-available`

[bramus]

This is a placeholder issue to discuss -webkit-fill-available. It is currently not covered in the test scenarios nor covered by an explainer.

[karlcow]

Do you mean covering stretch value in the tests?

width/height, (min|max)-(width|height) with the value stretch

• Firefox
  • Unprefix -moz-fit-content and -moz-available keywords. This is a Webcompat Priority P2 for Mozilla, as it creates webcompat issues.
  • support -moz-fit-content, and -moz-available for height/min-height/max-height (block axis)
  • [css-sizing][meta] implement CSS Intrinsic & Extrinsic Sizing Module Level 3
• Chromium
  • Behavior of width:-webkit-fill-available Chromium
  • [css-sizing] Unprefix fill/stretch-fit keyword when ready
• WebKit
  • Nothing specific but a list of related bugs
• CSS WG
  • CSS Box Sizing Module Level 4: stretch old value

It's a mess.

Two years ago I was wondering if there was a counter for -webkit-fill-available @foolip

[bramus]

• @emilio CSS Keyword, was supposed to only work on the inline axis. WebKit does something on the block axis, which authors now rely on.
• @emilio We don’t want this implement this right now.
• @emilio Useful for some viewport resizing behavior
• @emilio We could probably ask @bfgeek about it
• @dcrousso Will ask around as well

[bfgeek]

We've been slowly moving our -webkit-fill-available behaviour to the behaviour of what stretch should be.

The TL;DR of stretch is that is should act similarly[1] to 100% in standards mode. E.g. if the available-size in that dimension is definite.

(The way that I think about this at least is that auto maps onto either fit-content or stretch depending on the layout-mode and type of object. E.g. in flex in the cross-axis auto maps to stretch for example).

Our old implementation (and likely similar to current WebKit) behaved similarly to quirks-mode 100%. E.g. it would skip past any auto height elements in the containing-block chain up to the first "definite" thing. As a result of this lots of folks relied on this behaviour as it kinda-sorta behaved like 100dvh sometimes.

Blink can likely unprefix -webit-fill-available as stretch as we are reasonably confident that our -webkit-fill-available behaviour matches the spec.

Ian