-
Notifications
You must be signed in to change notification settings - Fork 747
Description
The text-orientation: upright
has the following text:
This value causes the used value of direction to be ltr, and for the purposes of bidi reordering, causes all characters to be treated as strong LTR.
This is a proposal to replace this text with notes for authors.
Background
In TPAC, we discussed test status and this is one of the "Required Tests That Do Not Meet Exit Criteria". @fantasai expressed that she is not comfortable to exit CR without passing this. Some discussions made there were:
- While this is not needed for joining RTL scripts such as Arabic, it does make sense for non-joining RTL scripts such as Hebrew.
- This is considered as extremely rare, the only use case we saw was building signage.
I then made further investigations as below to come to this proposal.
Workaround for authors
Authors can get almost the same effect by the following CSS:
.vertical-upright {
writing-mode: vertical-rl;
text-orientation: upright;
unicode-bidi: override;
direction: ltr;
}
Note, this workaround does not work if it has child elements with dir
attributes, or different values of unicode-bidi
properties.
Implementation status
- Blink crbug.com/663281, Edge 9713176, WebKit wkb.ug/164507 does not implement.
- Gecko 1315892 implements partially, by inserting the RLO (the same effect as the workaround above) if the block has
text-orientation: upright
.- It has the same restriction as above.
- It does not work when
text-orientation: upright
is applied to inline elements. - It does not change the used value of
direction
, and therefore other effects (e.g., logical properties, table cell orders, etc.) is not included.
Proposal
Given there is an easy workaround for authors, and given implementations are very limited, this proposal is to replace the text with a note how authors can get this effect.
I'm fine with other options, such as:
- Keep the text as is, by making the test failures as an exception to exit criteria.
- Make it "should" or "may".
but after reviewing the workaround and status, it looks to me that it's better to let authors know how to do it, rather than keeping the text and non-interoperable behavior.
Thoughts?