Strict Line Breaks Mode Should Respect Language-Specific Spacing Rules

Bug graveyard
1

Description

In Strict Line Breaks mode, Obsidian currently renders single line breaks (\n) as spaces uniformly across all languages. However, different languages have distinct typographic conventions regarding line breaks:

Language Type Example Languages Expected Line Break Behavior
Space-requiring English, French, German Line1\nLine2 → Line1 Line2
Non-space-requiring Chinese, Japanese 行1\n行2 → 行1行2

Current behavior forces space insertion even in languages where it’s typographically incorrect.

Steps to Reproduce

  1. Enable Strict Line Breaks in Settings → Editor
  2. Create test file with:
English
requires space

中文
不需要空格

Français
espace nécessaire

日本語
スペース不要
  1. Observe rendered output shows spaces after all line breaks

Expected Behavior

The renderer should:

  1. Detect language type (space-requiring vs. non-space-requiring)
  2. Apply appropriate line break rendering:
  • Space-requiring: Insert space ( )
  • Non-space-requiring: No space (direct concatenation)

Proposed Solution

  1. Language Detection:
  • Use Unicode ranges to identify language types:
    • CJK: U+4E00-U+9FFF, U+3040-U+30FF, etc.
    • Space-requiring: Latin (U+0000-U+007F extended)
  1. Render Logic:
function renderLineBreak(text) {
    const isCJK = /[\u4e00-\u9fff\u3040-\u30ff\uac00-\ud7af]/.test(text);
    return isCJK ? "" : " ";
}
2

Hello.

Can this be solved by using a zero-width space before each line break? Or does it still look wrong in the sample below?

==WITH== ZERO-WIDTH SPACE  
中文​
不需要空格

WITHOUT ZERO-WIDTH SPACE  
中文
不需要空格

==WITH== ZERO-WIDTH SPACE  
日本語​
スペース不要

WITHOUT ZERO-WIDTH SPACE  
日本語
スペース不要

zwspaces
zwspaces1450×798 98.4 KB

3

This is an interesting observation, however it isn’t not how markdown currently works (or specifically commonmark which we follow).

You can open a Feature Request to ask as to deviate from commonmark.
You may also be interested in this other FR.

4

Thank you for all you reply.
And I open a Feature Request here as a solution.