[csswg-drafts] [css-logical-1] values of `background-image-transform` and `border-image-transform` should be defined more precisely

dbaron has just created a new issue for https://github.com/w3c/csswg-drafts:

== [css-logical-1] values of `background-image-transform` and `border-image-transform` should be defined more precisely ==
The values of `background-image-transform` and `border-image-transform` should be defined more precisely in the [section on Logical Background and Border Images](https://drafts.csswg.org/css-logical-1/#background-and-borders).  I think the following issues need improvement, for both properties:
* `logical`'s definition says "The background images are transformed", but it doesn't say what assumption to start from.  I would assume that starting from the assumption of LTR inline flow and TTB block flow is the basis.
* `logical`'s definition says "to match to the flow-relative axis"... but there are two axes, and I think it means both.
* `logical`'s definition says "The values for the background-size property are flow-relative." but it doesn't say which value comes first.  I would assume inline-size is first followed by block-size, since that matches the current syntax for horizontal writing modes.
* `rotate`'s definition says that "inline base direction is ignored", but it doesn't say what to do instead.  I think it intends to say that the transformation is chosen such that the images is transformed such that what was originally the top becomes the block-start, what was originally the end becomes the block-end, and the inline edges are choses such that this transformation can be done as a rotation without a flip.
* The values of the `border-image-transform` values refer to the `border-image-*` properties, but it seems that (depending on the order of operations you imagine things happening in) the `border-image-slice` property may not be transformed since it describes offsets into the image prior to layout.

I'd also note that the fact that the logical property order and the physical property order match for horizontal RTL rather than horizontal LTR is particularly interesting in the context of `border-image-transfom: logical`, since it presumably (though this should really be specified) transforms the `border-image-width` and `border-image-outset` in this manner.

I got here from w3ctag/design-reviews#286.

Please view or discuss this issue at https://github.com/w3c/csswg-drafts/issues/2825 using your GitHub account

Received on Tuesday, 26 June 2018 05:37:37 UTC