[css-borders][css-backgrounds] Interaction of corner-shape with shape-outside needs to be determined

[smfr]

The proposed corner-shape property (#6521, #6980) means that box corners are not always rounded. This will require that https://drafts.csswg.org/css-backgrounds-3/#corner-shaping is updated other describe how to handle non-round corner shapes.

In addition, shape-outside references the <shape-box>, which may also not be a simple rounded rect with non-round corner-shapes. It's not super clear how a scoop corner shape affects the content box, for example.

[bfgeek]

If the use-counters are still low (they might be) it might be another reason to consider removing content-box/padding-box which would mitigate this type of question: #3872

[SebastianZ]

The proposed corner-shape property (#6521, #6980) means that box corners are not always rounded. This will require that https://drafts.csswg.org/css-backgrounds-3/#corner-shaping is updated other describe how to handle non-round corner shapes.

Note that corner-shape is already specified in CSS Borders 4. It still lacks an updated definition of how those shapes are generated. And I am planning to add some spec. text for that.

In addition, shape-outside references the <shape-box>, which may also not be a simple rounded rect with non-round corner-shapes. It's not super clear how a scoop corner shape affects the content box, for example.

The <shape-box> definition there obviously also needs to be adjusted to say that it follows the corner shape defined by border-radius and corner-shape.

Side note: It was resolved to drop notch and scoop for the time being, mainly due to implementation complexity. #6980 advocates to add them back by giving examples where authors would need them.

Sebastian

[smfr]

If a UA implements bevel, I think scoop and notch don't add much more complexity. In all cases, there are non-trivial adjustments of the shape of the padding box that are not just a case of adjusting padding box corners based on the size of the corner radius.