The Hidden Danger of Right-to-Left Decorative Characters in Web Typography
The footgun of right-to-left decorative characters
I discovered that using the Manichaean Punctuation Fleuron as a decorative separator caused unexpected text reordering due to its right-to-left bidirectional type. This Unicode quirk rearranges surrounding digits and text, but the issue is easily fixed by wrapping the character in a <bdi> tag or applying specific CSS to isolate its directionality.
Also they foolishly didn't allow post-deceased converts like some fast-growing religions.
- Georgelemental
If you are using plain text and don't have access to HTML or CSS markup, you can follow the RTL character with U+200E LEFT-TO-RIGHT MARK to achieve the same effect. And even if you do have access to those things, using U+200E ensures that operations that strip markup (like copy-paste) don't break your text.
- dhosek
Mixing R-L and L-R scripts (as has been discussed here on many occasions) is a ripe arena for mysterious behavior. Given that even monolingual texts in R-L scripts will often include L-R characters, it can get hairy quickly. A desire to try to avoid having visible markers in text around the transitions is partly why the Unicode spec around bidi text is so complicated.
- YokoZar
A fun hack I remember was that Windows internally used two left-to-right control characters in a window title as an indicator that the entire window should have its content flipped right to left, images and all (imagine a clock animation running counter-clockwise).
This meant that if you could control the window title somehow you could create some surprising behavior, such as putting both those characters at the beginning of a post on a web forum and then having both the web page and browser look backward after someone opened the link.
I believe most things filter for this now, of course.