I am trying to use Unicode variable names in g++, but it does not appear to work.
Does g++ not support Unicode variable names? Or is there some subset of Unicode (from which I'm not testing in)?
- ¤ g++ is just not standard-conforming wrt. characters in identifiers. But I don't know of any compiler that is conforming. It is my impression that most compilers limit the identifier characters to English A...Z and underscore, plus $ sign, which is wrong in two ways: not allowing the huge range of Unicode characters specified in Annex E of the standard (I've listed them at pastie.org/3110152), and allowing $, which the standard does not allow. In short, the standard and existing practice is very much at odds. Perhaps with C++11... ;-) Cheers & hth.,Cheers and hth. - Alf– Cheers and hth. - Alf2012-01-02 03:33:33 +00:00Commented Jan 2, 2012 at 3:33
- 1@Cheersandhth.-Alf Try clang :)Richard Smith– Richard Smith2013-07-24 07:24:02 +00:00Commented Jul 24, 2013 at 7:24
- Possible duplicate: 😃 (and other Unicode characters) in identifiers not allowed by g++.Peter Mortensen– Peter Mortensen2023-05-04 21:42:26 +00:00Commented May 4, 2023 at 21:42
- 2"it does not appear to work" Don't hesitate, go ahead and tell us how it doesn't work. We've been waiting for 13 years.n. m. could be an AI– n. m. could be an AI2023-05-08 22:13:26 +00:00Commented May 8, 2023 at 22:13
- @n. m. could be an AI: The OP has left the building: "Last seen more than 12 years ago"Peter Mortensen– Peter Mortensen2023-08-20 11:24:38 +00:00Commented Aug 20, 2023 at 11:24
Add a comment |
2 Answers
You have to specify the -fextended-identifiers flag when compiling. You also have to use \uXXXX or \uXXXXXXXX for Unicode (at least in GCC, it's Unicode).
Identifiers (variable/class names, etc.) in g++ can't be of UTF-8/UTF-16 or whatever encoding. They have to be:
identifier: nondigit identifier nondigit identifier digit A nondigit is
nondigit: one of universalcharactername _ a b c d e f g h i j k l m n o p q r s t u v w x y z A B C D E F G H I J K L M N O P Q R S T U V W X Y Z And a universalcharactername is
universalcharactername: \UXXXXXXXX \uXXXX Thus, if you save your source file as UTF-8, you cannot have a variable like:
int høyde = 10; It had to be written like:
int h\u00F8yde = 10; (which, in my opinion, would defeat the purpose. So just stick with a-z)
6 Comments
anon
Is there better support in clang?
Cheers and hth. - Alf
g++ is not standard-conforming here (but neither are other compilers, including Comeau). For standard C++, in the very first phase of translation "Any source file character not in the basic source character set (2.3) is replaced by the universal-character-name that designates that character", and the lexer rules operate on the result of that. In the C++11 standard this is specified in "Phases of translation" §2.2/1 1st list item.
Richard Smith
@anon Yes, clang allows accented characters in identifiers.
Sz.
9 years later G++ 9.1 is still blind to UTF-8 symbols, even with
-fextended-identifiers -finput-charset=UTF-8. (For reference, also MSVC++ does fine, either with -utf-8 or with a BOM in the source.) See also: stackoverflow.com/a/12693346/1479945 |
A one-line patch to the C++ preprocessor allows UTF-8 input. Details for GCC are given at UTF-8 Identifiers in GCC.
However, since the preprocessor is shared, the same patch should work for g++ as well. In particular, the patch needed, as of gcc-5.2 is
diff -cNr gcc-5.2.0/libcpp/charset.c gcc-5.2.0-ejo/libcpp/charset.c Output:
*** gcc-5.2.0/libcpp/charset.c Mon Jan 5 04:33:28 2015 --- gcc-5.2.0-ejo/libcpp/charset.c Wed Aug 12 14:34:23 2015 *************** *** 1711,1717 **** struct _cpp_strbuf to; unsigned char *buffer; ! input_cset = init_iconv_desc (pfile, SOURCE_CHARSET, input_charset); if (input_cset.func == convert_no_conversion) { to.text = input; --- 1711,1717 ---- struct _cpp_strbuf to; unsigned char *buffer; ! input_cset = init_iconv_desc (pfile, "C99", input_charset); if (input_cset.func == convert_no_conversion) { to.text = input; Note that for the above patch to work, a recent version of iconv needs to be installed that supports C99 conversions. Type iconv --list to verify this. Otherwise, you can install a new version of iconv along with GCC as described in the link above.
Change the configure command to
../gcc-5.2.0/configure -v --disable-multilib \ --with-libiconv-prefix=/usr/local/gcc-5.2 \ --prefix=/usr/local/gcc-5.2 \ --enable-languages="c,c++" if you are building for x86 and want to include the C++ compiler as well.
