使用版本控制，我的源代码或多或少可以用单个语法单位（句子、短语等）占据单个源代码行来完成，这很好。但这往往会导致（至少对我来说）源代码中有很多行以 M 破折号（---在源代码中）开头或结尾。
我发现，当发生这种情况时，换行符在文档中排版为空格。也就是说，M 破折号和它相邻的一个单词（但只有一个）之间有一个空格。也就是说，具体来说，下面的代码中数字 1 和 2 对我来说是错误的。如果您喜欢的话，第 4、5 和 6 行看起来不错，但只有第 3 行看起来符合我的要求。
\documentclass{article}
\begin{document}
1. A line in the source with an M-dash at the end---
like that one; with no space before it in the source.
2. A line in the source with an M-dash at the beginning
---like this one; also with no spaces around it.
3. A line---this one, I mean---with M-dashes in the middle,
and no spaces around them.
4. This line has an M-dash at the end of the line ---
like that; but with a space before it.
5. This one has the M-dash at the beginning of a line
--- like it? But there is a space after it.
6. This line --- the last line --- has M-dashes in the
middle, with spaces around them.
\end{document}
我非常希望：A) 不必在每次使用破折号时都求助于宏之类的笨拙的校对工具（“\sensiblemdash”或类似的东西）；B) 能够用破折号结束或开始行；C) 能够在源代码中使用看起来或多或少像破折号的东西；D) 不必在文档的排版中在破折号的两端都留有空格。（关于最后这一点，我知道这是某些人的偏好，但不是我的偏好，尽管我愿意接受争论。）我不介意在源代码的两侧输入空格，但我不希望它以这种方式排版。
编辑：忘了说：
$ latex --version
pdfTeX 3.14159265-2.6-1.40.17 (TeX Live 2016/Cygwin)
kpathsea version 6.2.2
答案1
源代码中的行尾与空格完全相同（它生成一个字符代码为 32 的空格标记，与空格字符生成的完全相同）
但是如果你输入破折号为 — 那么很容易定义忽略前后的空格，
\documentclass{article}
\usepackage[utf8]{inputenc}
\DeclareUnicodeCharacter{2014}{\leavevmode\unskip---\ignorespaces}
\begin{document}
1. A line in the source with an M-dash at the end—
like that one; with no space before it in the source.
2. A line in the source with an M-dash at the beginning
—like this one; also with no spaces around it.
3. A line—this one, I mean—with M-dashes in the middle,
and no spaces around them.
4. This line has an M-dash at the end of the line —
like that; but with a space before it.
5. This one has the M-dash at the beginning of a line
— like it? But there is a space after it.
6. This line — the last line — has M-dashes in the
middle, with spaces around them.
\end{document}