[Bug c/103446] New: Invalid wide multibyte character constant

[Bug c/103446] New: Invalid wide multibyte character constant

[zloten at mail dot ru]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=103446

            Bug ID: 103446
           Summary: Invalid wide multibyte character constant
           Product: gcc
           Version: 12.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: c
          Assignee: unassigned at gcc dot gnu.org
          Reporter: zloten at mail dot ru
  Target Milestone: ---

assert('Z' == 0x5A);
assert('Ы' == 0xD0AB);
assert('闩' == 0x00E997A9);
assert('𠂡' == 0xF0A082A1);
assert(U'Z' == 0x5A);
assert(U'Ы' == 0x042B);
assert(U'闩' == 0x95E9);
assert(U'𠂡' == 0x000200A1);
assert(L'Z' == 0x5A);
assert(L'Ы' == 0x042B);
assert(L'闩' == 0x95E9);
assert(L'𠂡' == 56481); ==> warning: character constant too long for its type   
         BUT MUST BE 0xDCA1D840

This problem occurs only with non-bmp wide supplementary characters.

[Bug c/103446] Invalid wide multibyte character constant

[zloten at mail dot ru]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=103446

--- Comment #1 from Zloten <zloten at mail dot ru> ---
56481 = 0xDCA1

[Bug preprocessor/103446] Invalid wide multibyte character constant

[pinskia at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=103446

--- Comment #2 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
Is there a full testcase including what options you used?
Did you use "-finput-charset=" and -fexec-charset= options?

[Bug preprocessor/103446] Invalid wide multibyte character constant

[zloten at mail dot ru]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=103446

--- Comment #3 from Zloten <zloten at mail dot ru> ---
No. Just - O2.

[Bug preprocessor/103446] Invalid wide multibyte character constant

[pinskia at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=103446

--- Comment #4 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
What target and what is the host?

[Bug preprocessor/103446] Invalid wide multibyte character constant

[zloten at mail dot ru]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=103446

--- Comment #5 from Zloten <zloten at mail dot ru> ---
I use the latest MinGW, target-host are Windows, x86-64.

[Bug preprocessor/103446] Invalid wide multibyte character constant

[pinskia at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=103446

Andrew Pinski <pinskia at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
         Resolution|---                         |INVALID
             Status|UNCONFIRMED                 |RESOLVED

--- Comment #6 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
Wide charater (wchar_t) on mingw is short (16bits) only.

That is why L'𠂡' does not fit into wchar_t.

[Bug preprocessor/103446] Invalid wide multibyte character constant

[zloten at mail dot ru]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=103446

--- Comment #7 from Zloten <zloten at mail dot ru> ---
It's very very strange. I've tested GCC 11.2.0(x86-64) and MinGW(x86-64). Both
have the same problem.
Let's do not use L suffix (it's implementation-defined). 
Let's use u suffix.
For both:

int test() {
    return   *(int*)u"𠂡"; // returns valid value 0xDCA1D840
}

int test2() {
   return u'𠂡'; // throws warning "character constant too long for its type"
                 // and returns invalid value 0xDCA1
}

Why we have no possibility to use non-BMP UTF-16 encoded value? 
It can be simply treated by compiler like a 32-bit integer constant.

Sorry for my bad English.

[Bug preprocessor/103446] Invalid wide multibyte character constant

[zloten at mail dot ru]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=103446

--- Comment #8 from Zloten <zloten at mail dot ru> ---
Sorry for my late reply