Is there a way to make g++ compile this program with Unicode identifiers?

Question

Consider:

int test(int )
{
   auto  = 5;
    += ;
   return ;
}

Is there a way to get g++ to compile this program? I've tried g++ -fextended-identifiers -finput-charset=utf-8 -c utf8-test.cpp and I get back errors like this:

utf8-test.cpp:1:9: error: stray ‘\360’ in program
    1 | int test(int )
      |         ^

It works great with clang.

Note that I'm not generally endorsing the practice of using emoji in identifiers. It was just a limit case for testing purposes. Though, I have a class it would be really nice to name ℤpField (with a nice type alias to ZPField to save everyone the pain of figuring out how to type it).

Edit: The TLDR answer appears to be that gcc will likely support this in gcc 10, but it doesn't right now, and this is a known issue. It's also not clear that this support is required by the standard and it's possible that you're required to use \U escaped hexadecimal or UCN sequences to specify Unicode characters in identifiers.

[This stackoverflow post](https://stackoverflow.com/a/2681996) says that g++ can not do this. — pschill, Mar 06 '20 at 18:08
It compiles with g++ https://wandbox.org/permlink/PqqOU3GarsYQ4V7B — anastaciu, Mar 06 '20 at 18:10
The emojis are out of the range of identifiers that are [allowed](https://en.cppreference.com/w/cpp/language/identifiers), although the ℤ is in the allowed range. — 1201ProgramAlarm, Mar 06 '20 at 18:12
Consider reading [this](https://gcc.gnu.org/bugzilla/show_bug.cgi?id=67224). GCC 10 will probably support this. It seems coherent with what @anastaciu said (as it uses GCC 10). — Jérôme Richard, Mar 06 '20 at 18:15
@anastaciu - Wandbox must be doing something a little strange with the preprocessor then. My version of g++ (`g++ (GCC) 9.2.1 20190827 (Red Hat 9.2.1-1) `) does not do this. — Omnifarious, Mar 06 '20 at 18:15
Between the comments and the answers at the question pointed to, I think this question has been answered. — Omnifarious, Mar 06 '20 at 18:16
@Omnifarious, mine too, it's most certainly the fact that wandbox has gcc HEAD 10.0.1 20200 version, when you compile it with 9.2.0 it also fails with consistent errors. — anastaciu, Mar 06 '20 at 18:20
reminded me of Stroustrups [`my_phone->off_hook();` proposal](http://www.stroustrup.com/whitespace98.pdf) — 463035818_is_not_an_ai, Mar 06 '20 at 18:47
A better duplicate may be *[ (and other Unicode characters) in identifiers not allowed by g++](https://stackoverflow.com/questions/12692067/)*. — Peter Mortensen, May 04 '23 at 21:42

Is there a way to make g++ compile this program with Unicode identifiers?

0 Answers0