P1949R7 C++ Identifier Syntax using Unicode Standard Annex 31

jensmaurer · jensmaurer · commit 53f9db65bc5a · 2021-06-08T21:01:28.000+02:00
diff --git a/source/back.tex b/source/back.tex
@@ -23,6 +23,12 @@ \chapter{Bibliography}
   \doccite{Unicode Text Segmentation} [online].
   Edited by Mark Davis. Revision 35; issued for Unicode 12.0.0. 2019-02-15 [viewed 2020-02-23].
   Available from: \url{http://www.unicode.org/reports/tr29/tr29-35.html}
+\item
+  The Unicode Consortium. Unicode Standard Annex, UAX \#31,
+  \doccite{Unicode Identifier and Pattern Syntax} [online].
+  Edited by Mark Davis. Revision 33; issued for Unicode 13.0.0.
+  2020-02-13 [viewed 2021-06-08].
+  Available from: \url{https://www.unicode.org/reports/tr31/tr31-33.html}
 \item
   IANA Time Zone Database.
   Available from: \url{https://www.iana.org/time-zones}
diff --git a/source/compatibility.tex b/source/compatibility.tex
@@ -1,15 +1,28 @@
 %!TEX root = std.tex
 \infannex{diff}{Compatibility}
 
-% TODO: Add this once we have differences.
-
-%\rSec1[diff.cpp20]{\Cpp{} and ISO \CppXX{}}
-%
-%\pnum
-%\indextext{summary!compatibility with ISO \CppXX{}}%
-%This subclause lists the differences between \Cpp{} and
-%ISO \CppXX{} (ISO/IEC 14882:2020, \doccite{Programming Languages --- \Cpp{}}),
-%by the chapters of this document.
+\rSec1[diff.cpp20]{\Cpp{} and ISO \CppXX{}}
+
+\rSec2[diff.cpp20.general]{General}
+
+\pnum
+\indextext{summary!compatibility with ISO \CppXX{}}%
+Subclause \ref{diff.cpp20} lists the differences between \Cpp{} and
+ISO \CppXX{} (ISO/IEC 14882:2020, \doccite{Programming Languages --- \Cpp{}}),
+by the chapters of this document.
+
+\rSec2[diff.cpp20.lex]{\ref{lex}: lexical conventions}
+
+\diffref{lex.name}
+\change
+Previously valid identifiers containing characters
+not present in UAX \#44 properties XID_Start or XID_Continue, or
+not in Normalization Form C, are now rejected.
+\rationale
+Prevent confusing characters in identifiers.
+Requiring normalization of names ensures consistent linker behavior.
+\effect
+Some identifiers are no longer well-formed.
 
 \rSec1[diff.cpp17]{\Cpp{} and ISO \CppXVII{}}
 
diff --git a/source/intro.tex b/source/intro.tex
@@ -70,6 +70,15 @@
 \end{footnote}
 Language Specification},
 Standard Ecma-262, third edition, 1999.
+\item
+The Unicode Consortium.
+Unicode Standard Annex, UAX \#44, \doccite{Unicode Character Database}.
+Edited by Ken Whistler and Lauren\c{t}iu Iancu.
+Available from: \url{http://www.unicode.org/reports/tr44/}
+\item
+The Unicode Consortium.
+The Unicode Standard, \doccite{Derived Core Properties}.
+Available from: \url{https://www.unicode.org/Public/UCD/latest/ucd/DerivedCoreProperties.txt}
 \end{itemize}
 
 \pnum
diff --git a/source/lex.tex b/source/lex.tex
@@ -325,6 +325,7 @@
     string-literal\br
     user-defined-string-literal\br
     preprocessing-op-or-punc\br
+    \textnormal{each} universal-character-name \textnormal{that cannot be one of the above}\br
     \textnormal{each non-whitespace character that cannot be one of the above}
 \end{bnf}
 
@@ -340,8 +341,12 @@
 (\grammarterm{import-keyword}, \grammarterm{module-keyword}, and \grammarterm{export-keyword}),
 identifiers, preprocessing numbers, character literals (including user-defined character
 literals), string literals (including user-defined string literals), preprocessing
-operators and punctuators, and single non-whitespace characters that do not lexically
-match the other preprocessing token categories. If a \tcode{'} or a \tcode{"} character
+operators and punctuators, and single \grammarterm{universal-character-name}s and non-whitespace characters that do not lexically
+match the other preprocessing token categories.
+If a single \grammarterm{universal-character-name}
+does not match any of the other preprocessing token categories,
+the program is ill-formed.
+If a \tcode{'} or a \tcode{"} character
 matches the last category, the behavior is undefined. Preprocessing tokens can be
 separated by
 \indextext{whitespace}%
@@ -602,8 +607,7 @@
 \nontermdef{pp-number}\br
     digit\br
     \terminal{.} digit\br
-    pp-number digit\br
-    pp-number identifier-nondigit\br
+    pp-number identifier-continue\br
     pp-number \terminal{'} digit\br
     pp-number \terminal{'} nondigit\br
     pp-number \terminal{e} sign\br
@@ -630,15 +634,21 @@
 \indextext{identifier|(}%
 \begin{bnf}
 \nontermdef{identifier}\br
-    identifier-nondigit\br
-    identifier identifier-nondigit\br
-    identifier digit
+    identifier-start\br
+    identifier identifier-continue\br
 \end{bnf}
 
 \begin{bnf}
-\nontermdef{identifier-nondigit}\br
+\nontermdef{identifier-start}\br
     nondigit\br
-    universal-character-name
+    universal-character-name \textnormal{of class XID_Start}
+\end{bnf}
+
+\begin{bnf}
+\nontermdef{identifier-continue}\br
+    digit\br
+    nondigit\br
+    universal-character-name \textnormal{of class XID_Continue}
 \end{bnf}
 
 \begin{bnf}
@@ -657,15 +667,8 @@
 \pnum
 \indextext{name!length of}%
 \indextext{name}%
-An identifier is an arbitrarily long sequence of letters and digits.
-Each \grammarterm{universal-character-name} in an identifier shall designate a
-character whose encoding in ISO/IEC 10646 falls into one of the ranges
-specified in \tref{lex.name.allowed}.
-The initial element shall not be a \grammarterm{universal-character-name}
-designating a character whose encoding falls into one of the ranges
-specified in \tref{lex.name.disallowed}.
-Upper- and lower-case letters are
-different. All characters are significant.
+The character classes XID_Start and XID_Continue
+are Derived Core Properties as described by UAX \#44.
 \begin{footnote}
 On systems in which linkers cannot accept extended
 characters, an encoding of the \grammarterm{universal-character-name} can be used in
@@ -674,69 +677,21 @@
 \tcode{\textbackslash u} in a \grammarterm{universal-character-name}. Extended
 characters can produce a long external identifier, but \Cpp{} does not
 place a translation limit on significant characters for external
-identifiers. In \Cpp{}, upper- and lower-case letters are considered
-different for all identifiers, including external identifiers.
+identifiers.
 \end{footnote}
-
-\begin{floattable}{Ranges of characters allowed}{lex.name.allowed}
-{lllll}
-\topline
-\tcode{00A8} &
-\tcode{00AA} &
-\tcode{00AD} &
-\tcode{00AF} &
-\tcode{00B2-00B5} \\
-\tcode{00B7-00BA} &
-\tcode{00BC-00BE} &
-\tcode{00C0-00D6} &
-\tcode{00D8-00F6} &
-\tcode{00F8-00FF} \\
-\tcode{0100-167F} &
-\tcode{1681-180D} &
-\tcode{180F-1FFF} &&\\
-\tcode{200B-200D} &
-\tcode{202A-202E} &
-\tcode{203F-2040} &
-\tcode{2054} &
-\tcode{2060-206F} \\
-\tcode{2070-218F} &
-\tcode{2460-24FF} &
-\tcode{2776-2793} &
-\tcode{2C00-2DFF} &
-\tcode{2E80-2FFF} \\
-\tcode{3004-3007} &
-\tcode{3021-302F} &
-\tcode{3031-D7FF} && \\
-\tcode{F900-FD3D} &
-\tcode{FD40-FDCF} &
-\tcode{FDF0-FE44} &
-\tcode{FE47-FFFD} & \\
-\tcode{10000-1FFFD} &
-\tcode{20000-2FFFD} &
-\tcode{30000-3FFFD} &
-\tcode{40000-4FFFD} &
-\tcode{50000-5FFFD} \\
-\tcode{60000-6FFFD} &
-\tcode{70000-7FFFD} &
-\tcode{80000-8FFFD} &
-\tcode{90000-9FFFD} &
-\tcode{A0000-AFFFD} \\
-\tcode{B0000-BFFFD} &
-\tcode{C0000-CFFFD} &
-\tcode{D0000-DFFFD} &
-\tcode{E0000-EFFFD} &
-\\
-\end{floattable}
-
-\begin{floattable}{Ranges of characters disallowed initially (combining characters)}{lex.name.disallowed}
-{llll}
-\topline
-\tcode{0300-036F} &
-% FIXME: Unicode v7 adds 1AB0-1AFF
-\tcode{1DC0-1DFF} &
-\tcode{20D0-20FF} &
-\tcode{FE20-FE2F} \\
-\end{floattable}
+The program is ill-formed
+if an \grammarterm{identifier} does not conform to
+Normalization Form C as specified in ISO/IEC 10646.
+\begin{note}
+Upper- and lower-case letters are considered different for all identifiers.
+\end{note}
+\begin{note}
+In translation phase 4,
+\grammarterm{identifier} also includes
+those \grammarterm{preprocessing-token}s\iref{lex.pptoken}
+differentiated as keywords\iref{lex.key}
+in the later translation phase 7\iref{lex.token}.
+\end{note}
 
 \pnum
 \indextext{\idxcode{import}}%
diff --git a/source/std.tex b/source/std.tex
@@ -146,6 +146,7 @@
 \include{limits}
 \include{compatibility}
 \include{future}
+\include{uax31}
 
 %%--------------------------------------------------
 %% back matter
diff --git a/source/uax31.tex b/source/uax31.tex
@@ -0,0 +1,131 @@
+%!TEX root = std.tex
+\infannex{uaxid}{Conformance with UAX \#31}
+
+\rSec1[uaxid.general]{General}
+
+\pnum
+This Annex describes the choices made in application of
+UAX \#31 (``Unicode Identifier and Pattern Syntax'')
+to \Cpp{} in terms of the requirements from UAX \#31 and
+how they do or do not apply to \Cpp{}.
+In terms of UAX \#31,
+\Cpp{} conforms by meeting the requirements
+R1 ``Default Identifiers'' and
+R4 ``Equivalent Normalized Identifiers''.
+The other requirements, also listed below,
+are either alternatives not taken or do not apply to \Cpp{}.
+
+\rSec1[uaxid.default]{R1 Default identifiers}
+
+\pnum
+UAX \#31 specifies a default syntax for identifiers
+based on properties from the Unicode Character Database, UAX \#44.
+The general syntax is
+\begin{codeblock}
+<Identifier> := <Start> <Continue>* (<Medial> <Continue>+)*
+\end{codeblock}
+where \tcode{<Start>} has the XID_Start property,
+\tcode{<Continue>} has the XID_Continue property, and
+\tcode{<Medial>} is a list of characters permitted between continue characters.
+For \Cpp{} we add the character U+005F, LOW LINE, or \tcode{_},
+to the set of permitted \tcode{<Start>} characters,
+the \tcode{<Medial>} set is empty, and
+the \tcode{<Continue>} characters are unmodified.
+In the grammar used in UAX \#31, this is
+\begin{codeblock}
+<Identifier> := <Start> <Continue>*
+<Start> := XID_Start + U+005F
+<Continue> := <Start> + XID_Continue
+\end{codeblock}
+
+\pnum
+This is described in the \Cpp{} grammar in \ref{lex.name},
+where \grammarterm{identifier} is formed from
+\grammarterm{identifier-start} or
+\grammarterm{identifier} followed by \grammarterm{identifier-continue}.
+
+\rSec2[uaxid.rfmt]{R1a Restricted format characters}
+
+\pnum
+If an implementation of UAX \#31 wishes to allow format characters
+such as ZERO WIDTH JOINER or ZERO WIDTH NON-JOINER
+it must define a profile allowing them, or
+describe precisely which combinations are permitted.
+
+\pnum
+\Cpp{} does not allow format characters in identifiers, so this does not apply.
+
+\rSec2[uaxid.stable]{R1b Stable identifiers}
+
+\pnum
+An implementation of UAX \#31 may choose to guarantee
+that identifiers are stable across versions of the Unicode Standard.
+Once a string qualifies as an identifier it does so in all future versions.
+
+\pnum
+\Cpp{} does not make this guarantee,
+except to the extent that UAX \#31 guarantees
+the stability of the XID_Start and XID_Continue properties.
+
+\rSec1[uaxid.immutable]{R2 Immutable identifiers}
+
+\pnum
+An implementation may choose to guarantee that
+the set of identifiers will never change
+by fixing the set of code points allowed in identifiers forever.
+
+\pnum
+\Cpp{} does not choose to make this guarantee.
+As scripts are added to Unicode,
+additional characters in those scripts may become available
+for use in identifiers.
+
+\rSec1[uaxid.pattern]{R3 Pattern_White_Space and Pattern_Syntax characters}
+
+\pnum
+UAX \#31 describes how languages that use or interpret patterns of characters,
+such as regular expressions or number formats,
+may describe that syntax with Unicode properties.
+
+\pnum
+\Cpp{} does not do this as part of the language,
+deferring to library components for such usage of patterns.
+This requirement does not apply to \Cpp{}.
+
+\rSec1[uaxid.eqn]{R4 Equivalent normalized identifiers}
+
+\pnum
+UAX \#31 requires that implementations describe
+how identifiers are compared and considered equivalent.
+
+\pnum
+\Cpp{} requires that identifiers be in Normalization Form C and
+therefore identifiers that compare the same under NFC are equivalent.
+This is described in \ref{lex.name}.
+
+\rSec1[uaxid.eqci]{R5 Equivalent case-insensitive identifiers}
+
+\pnum
+\Cpp{} considers case to be significant in identifier comparison, and
+does not do any case folding.
+This requirement does not apply to \Cpp{}.
+
+\rSec1[uaxid.filter]{R6 Filtered normalized identifiers}
+
+\pnum
+If any characters are excluded from normalization,
+UAX \#31 requires a precise specification of those exclusions.
+
+\pnum
+\Cpp{} does not make any such exclusions.
+
+\rSec1[uaxid.filterci]{R7 Filtered case-insensitive identifiers}
+
+\pnum
+\Cpp{} identifiers are case sensitive, and
+therefore this requirement does not apply.
+
+\rSec1[uaxid.hashtag]{R8 Hashtag identifiers}
+
+\pnum
+There are no hashtags in \Cpp{}, so this requirement does not apply.
