[css-viewport] Behavior of segments when there is only 1 segment

[bramus]

The spec currently reads the following for the segments property of window.viewport

4. Returns null if there is only a single viewport segment and abort these steps.

This very counterintuitive when trying things out and I would have expected for 1 segment to reported, with its values at origin 0,0 and its width/height equal to the innerWidth and innerHeight respectively.

Admittedly the fallback authors can do is pretty easy here, yet I still find it weird to return null as currently specced. That keyword is typically reserved to indicate that there is no value or there is no such thing, whereas there is definitely 1 segment here.

Having it return 1 segment would also help distinguish the 1 segment situations vs “the document is currently not active” situation. The latter also returns null, as per spec:

1. If the Viewport’s associated Document is not fully active, return null

In summary, the proposed change is to return 1 segment instead of null when there is only 1 segment.

/cc @darktears

[darktears]

This has been discussed in the past unfortunately I can't find the discussion anymore.

Semantically a segment is a part of something that is divided. If the viewport is not divided, then there are no segments. When the device is not folded there are no division therefore no segments.

We don't want authors to write code against one segment because they should treat that case as the same as if there wasn't any split (meaning the device is used flat in the case of a foldable) or it's just a regular device. Furthermore, that one segment will return the same information as the viewport size bringing no useful information. We also don't want developer to feature check on this.

Finally, not returning a single segment match the CSS part of this API where the MQs do not resolve for a value of one segment. In other words, MQs values are always >= 2 for the segments.

[bramus]

We don't want authors to write code against one segment because they should treat that case as the same as if there wasn't any split (meaning the device is used flat in the case of a foldable) or it's just a regular device.

It’s not about writing code against 1 segment, but about consistency when enumerating the segments. We keep saying “1 segment” but that does not rhyme with null.

Finally, not returning a single segment match the CSS part of this API where the MQs do not resolve for a value of one segment.

Which MQ are you talking about? AFAIK there is only the device-posture media feature and some env-vars that people can use.

---

I asked on social media (bsky, masto) what developers think about this:

📊 Poll time: say there is an API to enumerate how many segments the viewport has. E.g. for a foldable phone the number would be 2 when the device is unfolded.

What would you expect this API to return when the foldable is closed, or when it’s a non-foldable device?

A. null
B. 1
C. ???

There was an overwhelming for answer B: 1 segment.

[darktears]

It’s not about writing code against 1 segment, but about consistency when enumerating the segments. We keep saying “1 segment” but that does not rhyme with null.

That's the thing, I don't want to talk about "1 segment" ever because it doesn't make sense conceptually IMHO.

Which MQ are you talking about? AFAIK there is only the device-posture media feature and some env-vars that people can use.

https://www.w3.org/TR/mediaqueries-5/#mf-horizontal-viewport-segments

I asked on social media (bsky, masto) what developers think about this:

📊 Poll time: say there is an API to enumerate how many segments the viewport has. E.g. for a foldable phone the number would be 2 when the device is unfolded.
What would you expect this API to return when the foldable is closed, or when it’s a non-foldable device?
A. null
B. 1
C. ???

There was an overwhelming for answer B: 1 segment.

Well, I guess I'll need to update the implementation and the spec.

[darktears]

In summary based on the various discussions on social websites we would be updating the specification to:

• Return 1 segment when the device is used flat or any other non-foldable device.
• The unique segment returned will be the same size as the viewport.

Other cases will behave the same as today, returning n segments depending on the device state and capabilities (one or multiple folds).

[bramus]

Slight tweak of the first proposed resolution, to match the PR’d spec change: “Return an array with just the 1 segment when the device is used flat or any other non-foldable device.”

[astearns]

The CSSWG will automatically accept this resolution one week from now if no objections are raised here. Anyone can add an emoji to this comment to express support. If you do not support this resolution, please add a new comment.

Proposed Resolution: Return an array with just one segment when the device is used flat or any other non-foldable device.

[astearns]

RESOLVED: Return an array with just one segment when the device is used flat or any other non-foldable device.

[darktears]

Spec change landed and WPT test added