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

--- Comment #20 from Jonathan Clark <[email protected]> ---
In ODF (and LO), paragraph directions are explicitly defined in terms of the
Unicode BiDi algorithm, and are provided for the purposes of languages we
conventionally consider RTL. The right-to-leftish paragraph writing modes are
appropriate for these conventional RTL use cases. They are neither intended for
nor defined in a way that is appropriate for other uses. (Reference: ODF
Appendix E.1.)

In the exceptionally rare case that you want CJK text with right-to-left
character progression, setting your paragraph to RTL will not do what you want
because Unicode defines CJK characters as bidi class L. In order to support
such right-to-left CJK, we would need to add some new feature. How to design
that feature would be a separate discussion. Right-to-left CJK is not obviously
relevant to this specific discussion about an existing LO feature which caters
to conventional RTL use cases.

This is also naturally so. Consider that English may also be written
right-to-left. In this case, characters are mirrored. Like the Japanese
example, you will mostly see this on signs and vehicles. Roughly 20% of
humanity speaks English. Despite these facts, we would not argue that
right-to-left paragraphs should cater to English and display English text
mirrored, because we intuitively understand this is not the purpose of
right-to-left paragraphs. Right-to-left paragraphs are for languages like
Arabic and Hebrew. When we make a paragraph right-to-left, we expect English
and Japanese characters inside that paragraph to be laid out according to what
makes sense for Arabic and Hebrew readers, not according to what makes sense
for English or Japanese readers.

Right-to-left horizontal CJK is also a niche use case. Claiming it affects 10%
of humanity distorts the issue. In practice, 0% of humanity needs to regularly
write right-to-left horizontal CJK, and the 0% plus epsilon of humanity with
this occasional need may already accomplish it by other means. Arguments based
on total user impact can be helpful, but we should try to make meaningful
estimates rather than exaggerations.

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

Reply via email to