+2026-04-24 Paul Eggert <eggert@cs.ucla.edu>
+
+ uchar-h: <string.h> etc. namespace cleanup
+ This is only a partial cleanup; to be cleaner we’d need to
+ move declarations of Gnulib extensions like c32isalpha
+ into a separate .h file. However, if no Gnulib modules that
+ extend <uchar.h> are used, Gnulib <uchar.h> is now pretty clean on
+ recent GNUish platforms.
+ * lib/uchar.in.h: On GNUish platforms, include <stdint.h>,
+ <wchar.h>, <wctype.h> only if needed. Do not include <string.h>,
+ as we never need it directly: even if we use <string.h>’s memset
+ via mbszero, <wchar.h> should include <string.h> if needed, as
+ <wchar.h> defines mbszero. Move a static_assert from here to tests,
+ as the static_assert uses a symbol that is no longer guaranteed
+ to be visible.
+ * modules/uchar-h (Depends-on): Do not depend on assert-h.
+ * tests/test-uchar-h.c: Move a static_assert here from lib/uchar.in.h,
+ and include <wchar.h> so that wchar_t is guaranteed to be visible.
+
2026-04-24 Bruno Haible <bruno@clisp.org>
version-etc: Optionally emit another line for the man page.
POSIX specification:@* @url{https://pubs.opengroup.org/onlinepubs/9799919799/basedefs/uchar.h.html}
-Defines the types @code{char16_t}, @code{char32_t} and declares the
+Defines the types @code{char16_t}, @code{char32_t},
+@code{mbstate_t} and @code{size_t}, and declares the
functions @code{mbrtoc16}, @code{c16rtomb}, @code{mbrtoc32},
@code{c32rtomb}.
implementations.
@end itemize
+Portability problems fixed by Gnulib modules
+@code{btoc32},
+@code{c32_apply_mapping}, @code{c32_apply_type_test},
+@code{c32_get_mapping}, @code{c32_get_type_test},
+@code{c32isalnum}, @code{c32isalpha}, @code{c32isblank},
+@code{c32iscntrl}, @code{c32isdigit}, @code{c32isgraph},
+@code{c32islower}, @code{c32isprint}, @code{c32ispunct},
+@code{c32isspace}, @code{c32isupper}, @code{c32isxdigit},
+@code{c32rtomb}, @code{c32snrtombs}, @code{c32srtombs},
+@code{c32stombs}, @code{c32swidth}, @code{c32tob},
+@code{c32tolower}, @code{c32toupper}, @code{c32width}:
+@itemize
+@item
+The @code{char32_t} type of ISO C 23 and POSIX @code{<uchar.h>} lacks
+counterparts of several standard @code{wchar_t} functions declared in
+@code{<wchar.h>} and @code{<wctype.h>}.
+This shortcoming is ameliorated by these Gnulib modules,
+each of which arranges for @code{<uchar.h>} to declare a
+Gnulib-specific function and to make visible the symbols @code{wint_t}
+and @code{WEOF} if the symbols are part of that function's API.
+@end itemize
+
Portability problems not fixed by Gnulib:
@itemize
@item
#error "Please include config.h first."
#endif
-/* Get uint_least16_t, uint_least32_t. */
-#include <stdint.h>
-
-/* Get mbstate_t, size_t. */
-#include <wchar.h>
-
-/* For the inline functions. */
-#include <string.h>
-#include <wctype.h>
+/* If needed, get uint_least16_t, uint_least32_t.
+ Although POSIX allows <uchar.h> to make all symbols visible from <stdint.h>,
+ our includers should not rely on this. */
+#if (! (defined __cplusplus \
+ ? @CXX_HAVE_UCHAR_H@ || @CXX_HAS_UCHAR_TYPES@ \
+ : @HAVE_UCHAR_H@) \
+ || @GNULIBHEADERS_OVERRIDE_CHAR16_T@ || @GNULIBHEADERS_OVERRIDE_CHAR32_T@)
+# include <stdint.h>
+#endif
+
+/* If needed, get btowc, mbstate_t, mbszero, size_t, wctob, wint_t, WEOF.
+ Although POSIX allows <uchar.h> to make all symbols visible from <wchar.h>,
+ our includers should not rely on this, except they can rely on wint_t and WEOF
+ when part of the API of a Gnulib module extending <uchar.h> that needs
+ these two symbols. */
+#if (! (/* The underlying <uchar.h> defines mbstate_t, size_t. */ \
+ defined __cplusplus ? @CXX_HAVE_UCHAR_H@ : @HAVE_UCHAR_H@) \
+ || (/* These need wint_t and maybe WEOF and a <wchar.h> function. */ \
+ @GNULIB_BTOC32@ || @GNULIB_C32TOB@) \
+ || (/* These need mbszero. */ \
+ (@GNULIB_C32STOMBS@ || @GNULIB_MBSTOC32S@)))
+# include <wchar.h>
+#endif
+
+/* If needed, get iswalnum, iswalpha, iswblank, iswcntrl, iswctype,
+ iswdigit, iswgraph, iswlower, iswprint, iswpunct, iswspace,
+ iswupper, iswxdigit, towctrans, towlower, towupper, wctrans,
+ wctrans_t, wctype, wctype_t, wint_t, WEOF.
+ Our includers should not rely on this, except they can rely on wint_t and WEOF
+ when part of the API of a Gnulib module extending <uchar.h> that needs
+ these two symbols. */
+#if (/* These need wint_t and maybe wctrans_t, wctype_t, WEOF, \
+ and a <wchar.h> function. */ 0 \
+ || @GNULIB_C32ISALNUM@ || @GNULIB_C32ISALPHA@ || @GNULIB_C32ISBLANK@ \
+ || @GNULIB_C32ISCNTRL@ || @GNULIB_C32ISDIGIT@ || @GNULIB_C32ISGRAPH@ \
+ || @GNULIB_C32ISLOWER@ || @GNULIB_C32ISPRINT@ || @GNULIB_C32ISPUNCT@ \
+ || @GNULIB_C32ISSPACE@ || @GNULIB_C32ISUPPER@ || @GNULIB_C32ISXDIGIT@ \
+ || @GNULIB_C32TOLOWER@ || @GNULIB_C32TOUPPER@ \
+ || @GNULIB_C32_APPLY_MAPPING@ || @GNULIB_C32_APPLY_TYPE_TEST@ \
+ || @GNULIB_C32_GET_MAPPING@ || @GNULIB_C32_GET_TYPE_TEST@)
+# include <wctype.h>
+#endif
/* The __attribute__ feature is available in gcc versions 2.5 and later.
The attribute __pure__ was added in gcc 2.96. */
#if !(defined __cplusplus ? @CXX_HAVE_UCHAR_H@ || @CXX_HAS_CHAR8_TYPE@ : @HAVE_UCHAR_H@)
/* An 8-bit variant of wchar_t.
- Note: This type is only mandated by ISO C 23 or newer, and it does
- denote UTF-8 units. */
+ Note: This type is mandated by ISO C 23 or newer, and denotes UTF-8 units. */
typedef unsigned char char8_t;
#elif @GNULIBHEADERS_OVERRIDE_CHAR8_T@
#if !(defined __cplusplus ? @CXX_HAVE_UCHAR_H@ || @CXX_HAS_UCHAR_TYPES@ : @HAVE_UCHAR_H@)
/* A 16-bit variant of wchar_t.
- Note: This type is only mandated by ISO C 11 or newer. In ISO C 23
+ Note: This type is mandated by ISO C 11 or newer. In ISO C 23
and newer, it denotes UTF-16 units; in older versions of ISO C it did
- so only on platforms on which __STDC_UTF_16__ was defined. */
+ so on platforms on which __STDC_UTF_16__ was defined. */
typedef uint_least16_t char16_t;
#elif @GNULIBHEADERS_OVERRIDE_CHAR16_T@
#if !(defined __cplusplus ? @CXX_HAVE_UCHAR_H@ || @CXX_HAS_UCHAR_TYPES@ : @HAVE_UCHAR_H@)
/* A 32-bit variant of wchar_t.
- Note: This type is only mandated by ISO C 11 or newer. In ISO C 23
+ Note: This type is mandated by ISO C 11 or newer. In ISO C 23
and newer, it denotes UTF-32 code points; in older versions of ISO C
- it did so only on platforms on which __STDC_UTF_32__ was defined.
+ it did so on platforms on which __STDC_UTF_32__ was defined.
In gnulib, we guarantee that it denotes UTF-32 code points if and
only if the module 'uchar-h-c23' is in use. */
typedef uint_least32_t char32_t;
- because GCC >= 4.9 defines these macros on all platforms, even on
FreeBSD and Solaris.
We should better not use __STD_UTF_16__, __STD_UTF_32__ either, because
- these macros are misspellings, only defined by Android's <uchar.h>. */
+ these macros are misspellings, defined only by Android's <uchar.h>. */
#if defined __STDC_ISO_10646__ && !_GL_SMALL_WCHAR_T
/* glibc, musl libc */
# define _GL_WCHAR_T_IS_UCS4 1
#endif
-#if _GL_WCHAR_T_IS_UCS4
-static_assert (sizeof (char32_t) == sizeof (wchar_t));
-#endif
-/* Convert a single-byte character to a 32-bit wide character. */
+/* Convert a single-byte character C to a 32-bit wide character,
+ or to WEOF if C is invalid. */
#if @GNULIB_BTOC32@
# if _GL_WCHAR_T_IS_UCS4 && !defined IN_BTOC32
_GL_BEGIN_C_LINKAGE
#endif
-/* Converts a 32-bit wide character to a multibyte character. */
+/* Convert a 32-bit wide character to a multibyte character. */
#if @GNULIB_C32RTOMB@
# if @REPLACE_C32RTOMB@
# if !(defined __cplusplus && defined GNULIB_NAMESPACE)
#endif
-/* Converts a 32-bit wide character to unibyte character.
- Returns the single-byte representation of WC if it exists,
+/* Convert a 32-bit wide character to unibyte character.
+ Return the single-byte representation of WC if it exists,
or EOF otherwise. */
#if @GNULIB_C32TOB@
# if _GL_WCHAR_T_IS_UCS4 && !defined IN_C32TOB
#endif
-/* Converts a multibyte character to a 32-bit wide character. */
+/* Convert a multibyte character to a 32-bit wide character. */
#if @GNULIB_MBRTOC32@
# if @REPLACE_MBRTOC32@
# if !(defined __cplusplus && defined GNULIB_NAMESPACE)
#endif
-/* Converts a multibyte character and returns the next 16-bit wide
+/* Convert a multibyte character and returns the next 16-bit wide
character. */
#if @GNULIB_MBRTOC16@
# if @REPLACE_MBRTOC16@
projects / thirdparty / gnulib.git / commitdiff
