alreq
alreq copied to clipboard
w3c
Recommended symbols/signs for Paragraph and Section & general mirroring gaps
Section sign (§) and Pilcrow (¶) are commonly used in Latin-script documents as symbols for Section and Paragraph, usually used for linking (in plain text on paper, or hyper link on web/ebook) to internal/external sections/paragraphs.
Pilcrow (¶) is also commonly used in "show control character" mode of Word Processors as a symbol for "new line", along side the U+23CE RETURN SYMBOL (⏎).
None of these are that common in Arabic-script documents/apps. We need to research on common/traditional practices for the section/paragraph symbols.
About the RETURN SYMBOL, we need to see if there has been any keyboards using a mirrored-version of the symbol for denoting NL/CR.
See Also:
- https://en.wikipedia.org/wiki/Section_sign
- https://en.wikipedia.org/wiki/Pilcrow
- https://en.wikipedia.org/wiki/Carriage_return
Actually, i'm surprised that Unicode doesn't treat ⏎ [U+23CE RETURN SYMBOL] as a mirrored character. Perhaps we should ask them about it.
I would be surprised to see bidi mirroring for the Return Symbol. Most of the mirrored characters do have another codepoint assigned for the mirrored glyph, and that's been mostly the rule of thumb to consider giving codepoints bidi mirroring property.
I don't think U+23CE RETURN SYMBOL has had a mirrored glyph encoded in any charset or font. Although, I'm guessing that there could be cases that people have gone into the trouble of making one, so they can document the software behavior better.
I'm wondering what was the solution in bidi-enabled word processors of the early 90's, like Zarnegar and the Perso-Arabic enabled Mac. IIRC, MS Word 6.0+ has always used the same glyphs as English/Latin for other languages/scripts/directions.
@r12a, It's even more curious because other keyboard chars/keys (Tab, BS) exist in right and left codes
⇤ LEFTWARDS ARROW TO BAR U+21E4 ⇥ RIGHTWARDS ARROW TO BAR U+21E5 ⌦ ERASE TO THE RIGHT U+2326 ⌫ ERASE TO THE LEFT U+232B
but not ⏎ RETURN SYMBOL U+23CE
BTW, "mirrored" is a property related to bidi (https://www.unicode.org/glossary/#mirrored_property).
Some chars "({[<⪡..." are automatically mirrored giving ")}]>⪢ ..." in LTR/RTL modes.
Some other characters, e.g. based on arrows like ↩ (U+21A9), have mirrored images but don't flip to ↪ (U+21AA) in bidi texts. (Why?)
Most of the mirrored characters do have another codepoint assigned for the mirrored glyph, and that's been mostly the rule of thumb to consider giving codepoints bidi mirroring property.
Looking at https://unicode.org/cldr/utility/list-unicodeset.jsp?a=[:Bidi_Mirrored=Yes:] it seems to me that quite a high number of mirrored characters are unpaired. They are mostly mathematical. I suppose extending this to symbols like return might open the floodgates for all sorts of other symbols and pictograms, etc.
A question is which mirrored chars should be flipped in RTL and which not? Browsers have different opinions. Some examples : 1- Some are flipped (N-ary Summation, U+2211) some not (Contains as Member, U+220B) a ∑ b ا ∑ ب
a ∋ b ا ∋ ب
2- Cube/Forth Root (U+221B, U+221C) flipped (bug? browsers overzealous?) a ∛ b ا ∛ ب
3- On the other hand, Square Root (U+221A) not flipped in some browsers a √ b ا √ ب
A screen capture:
For these math symbols I think Firefox will mirror the glyphs only if the font has mirrored glyphs for them in rtlm feature since there is no mirrored encoded characters for these, try using XITS Math font.
Other browsers seem to be just flipping the glyphs, but this is the wrong thing to do since some mirrored symbols are not completely flipped (e.g. in ∳ the integral part is flipped but the arrow should not).
Regarding the Bidi Mirroring discussion, see also:
- http://www.unicode.org/L2/L2018/18049-bidi-mirror.txt
- http://www.unicode.org/L2/L2017/17438-bidi-math-fdbk.html
FYI: A new revision of the first document is accepted and the data file will be updated accordingly:
- https://www.unicode.org/L2/L2018/18049r-bidi-mirror.txt