[tkoeppe]

[eisenwave]

I feel like this would be cleaner if you used the \item-on-separate-line form here:

Suggested change

	\begin{itemize}
	\item header names;
	\item placeholder tokens produced by preprocessing \tcode{import}
	and \tcode{module} directives
	(\grammarterm{import-keyword}, \grammarterm{module-keyword},
	and \grammarterm{export-keyword});
	\item identifiers;
	\item preprocessing numbers;
	\item character literals (including user-defined character literals);
	\item string literals (including user-defined string literals);
	\item preprocessing operators and punctuators;
	\item single non-whitespace characters that do not lexically
	match the other preprocessing token categories.
	\end{itemize}
	\begin{itemize}
	\item
	header names;
	\item
	placeholder tokens produced by preprocessing \tcode{import} and \tcode{module} directives
	(\grammarterm{import-keyword}, \grammarterm{module-keyword}, and \grammarterm{export-keyword});
	\item
	identifiers;
	\item
	preprocessing numbers;
	\item
	character literals
	(including user-defined character literals);
	\item
	string literals
	(including user-defined string literals);
	\item
	preprocessing operators and punctuators;
	\item
	single non-whitespace characters
	that do not lexically match the other preprocessing token categories.
	\end{itemize}

It's just more maintainable that way and the semantic line breaks come out cleaner. Perhaps we could also add some forward references here. We mention "string literals" which are described in subsequent subclause [lex.string], but don't provide a forward reference. That would make each item longer as well and justify having multiple lines for more \items.