Maybe these two fragments of the inline box are still considered one inline box fragmented between two anonymous block-level fragmentainers, similarly to how the fragments of inline blocks split into different line boxes belonging to different column/page boxes are still considered one (though fragmented) inline box?

I suppose that fragmentation of inline boxes between different line boxes (not limited to this particular case) should be clarified in the Fragmentation spec. Probably we need another issue about getting this type of fragmentation described properly?

@SelenIT My understanding from #1477 is that the inline element ends up generating multiple boxes. Generating a single box with multiple fragments would solve this, but then I'm not sure the box tree would be much well-defined, as I asked in #1477.

I commented #1477 with my reasoning why I tend to see this as a single but fragmented box. I support the idea from your initial post there that this should be explained in CSS Break, because there are many other cases where inline boxed behave very similarly.

(BTW, I'm not very comfortable with the "inline box that can be broken between line boxes" concept at all. For me, the "box" is more like a single rectangle with definite position and dimensions that clearly define the place it takes on the screen. It's hard to me to think of an inline element broken between several lines as a single "box", it feels more like a set/collection of "boxes", each with its own dimensions and position. Maybe it's just me? But having "box fragment" term for each of these single rectangles helps me a bit:)

BTW, I'm not very comfortable with the "inline box that can be broken between line boxes" concept at all. For me, the "box" is more like a single rectangle with definite position and dimensions that clearly define the place it takes on the screen.

In CSS terminology, "box" is a pre-layout concept; boxes don't have positions or sizes at all. "Fragment" is what results from the layout process, and has size and position.

I raised the block-splits-inline question (splits box, or splits fragments?) in #1477; this issue depends on that one.