[css-overflow] Questions regarding breaks in ellipsis strings

[xiaochengh]

1. Regarding block-ellipsis, the current spec says "The UA must treat the block overflow ellipsis as an unbreakable string" [1]. Is it referring to line breaking only, or also bidi re-ordering and fragmentation? I don't see anything wrong with allowing bidi reordering in the ellipsis string.

2. Regarding text-overflow, the current spec text doesn't mention anything about breaks. Should we add the same requirement as block-ellipsis, namely, no linebreaking is allowed? Currently, only Firefox supports custom ellipsis strings, and FF ignores forced breaks there.

[1] https://www.w3.org/TR/css-overflow-4/#block-ellipsis

[tabatkins]

1. I think you're right, it's just line breaking. However, I suspect the ellipsis should be treated as bidi isolated, so it doesn't reorder with the preceding inline content. @fantasai ?
2. Yeah, unless there's some compat requirement, I expect the two ellipsises should work the same.

[frivoal]

However, I suspect the ellipsis should be treated as bidi isolated, so it doesn't reorder with the preceding inline content

The spec says:

This inline is assigned unicode-bidi: plaintext

unicode-bidi: plaintext does cause bidi isolation, so that's covered already.

Yeah, unless there's some compat requirement, I expect the two ellipsises should work the same.

I am not sure this is a good assumption to make, in general. text-overflow is meant to be a paining effect, which can handled the overflowing content being scrolled in and out of view. block ellipsis is meant to integrate with line breaks. There may be some overlap in some behavior, and gratuitous divergence is not desirable, but the constraints are different.

That said, the definition of test-overflow: ellipsis does say:

Ellipsing only affects rendering and must not affect layout

There are various other aspects of the spec which hint that test-overflow: ellipsis is intended as a paint-time effect, and so there cannot possibly be any interaction with layout, bidi, fragmentation…

The Chrome implementation is known to diverge from that spec somewhat, and I think it is done with some degree of integration into layout. That has yet to be raised against the spec or resolved, but I don't expect it would (deliberately) be different about isolating bidi reordering in the string from bidi-reordering on the rest of the line.

@xiaochengh, does that answer your question?

We might want to add a small clarification in the spec(s) that bidi-reordering within the ellipsis is allowed.

[frivoal]

Effectively, I think these strings behave as if they were using white-space: preserve-spaces nowrap