https://bugs.documentfoundation.org/show_bug.cgi?id=168537

--- Comment #21 from Khaled Hosny <[email protected]> ---
In addition to what Jonathan said:

We need to distinguish between paragraph direction (which is the issue here)
and character-level direction (which is an orthogonal issue).

In Unicode BiDi, paragraph direction is left to higher level protocols (and
there is a suggested heuristic for when such protocols are missing) because it
is something only the author of the text knows for certain and it can’t be
reliably deduced from the text itself. That is why I say direction is content
not formatting. It is similar to language, if I set the language of a piece of
text to Arabic or French, that is part of the content not something I consider
formatting. Paragraph direction is the same.

Character-level direction is different, because it can be reliably deduced from
character properties (there can still be ambiguities, but Unicode has BiDi
control characters for these, because when there is an ambiguity, solving it
becomes part of the content). Some scripts have fluid directionality (CJK is
not unique in this, many historical scripts also had such fluidity). But
paragraph direction has no bearing to this. The way to make such text RTL (if
it is LTR in Unicode) is to use Unicode direction overrides, RLO and LRO. In
HTML there is also <bdi>
(https://developer.mozilla.org/en-US/docs/Web/HTML/Reference/Elements/bdi) for
isolates and <bdo>
(https://developer.mozilla.org/en-US/docs/Web/HTML/Reference/Elements/bdo) for
overrides, LibreOffice can benefit from a similar functionality, but that is a
different issue of what is being discussed here.

-- 
You are receiving this mail because:
You are the assignee for the bug.

Reply via email to