Issue #1121 new

C++ literal errors

Anonymous created an issue
// integer literals:
234;        // decimal, OK
0234;       // octal, OK
0x1234abCD; // hexadecimal, OK
0X1234abCD; // hexadecimal, fail
0b1010;     // binary, fail
0B1010;     // binary, fail

1;        // OK
1u;       // OK
1U;       // OK
1l;       // OK
1L;       // OK
1ll;      // OK
1LL;      // OK
1ul;      // OK
1Ul;      // OK
1uL;      // OK
1UL;      // OK
1ull;     // OK
1Ull;     // OK
1uLL;     // OK
1ULL;     // OK

01;        // OK
01u;       // OK
01U;       // OK
01l;       // OK
01L;       // OK
01ll;      // OK
01LL;      // OK
01ul;      // OK
01Ul;      // OK
01uL;      // OK
01UL;      // OK
01ull;     // OK
01Ull;     // OK
01uLL;     // OK
01ULL;     // OK

0x1;        // OK
0x1u;       // OK
0x1U;       // OK
0x1l;       // OK
0x1L;       // OK
0x1ll;      // OK
0x1LL;      // OK
0x1ul;      // OK
0x1Ul;      // OK
0x1uL;      // OK
0x1UL;      // OK
0x1ull;     // OK
0x1Ull;     // OK
0x1uLL;     // OK
0x1ULL;     // OK

0b1;        // fail
0b1u;       // fail
0b1U;       // fail
0b1l;       // fail
0b1L;       // fail
0b1ll;      // fail
0b1LL;      // fail
0b1ul;      // fail
0b1Ul;      // fail
0b1uL;      // fail
0b1UL;      // fail
0b1ull;     // fail
0b1Ull;     // fail
0b1uLL;     // fail
0b1ULL;     // fail

// integer literals with ' as separator always fail:
1'123'456'789;    // same as 1234567890
0123'456;         // same as 0123456
0xABCD'1234;      // same as 0xABCD1234
0XABCD'1234;      // same as 0XABCD1234
0b1111'0000'0000; // same as 0b111100000000
0B1111'0000'0000; // same as 0B111100000000
// the ' can appear as separator on every position of an integer but for direct
// after a prefix (0, 0b, 0B, 0x or 0X), after another ' ('' is not allowed)
// and of course at the begin and after the last _digit_ (not suffix) of an integer literal

978'3'446'43981'8; // is a correct integer literal, same as 9783446439818
// (it is an ISBN, as example for a not equidistant separation use case)

// floating point literals:
4.24;    // OK
4.24f;   // OK
4.24F;   // OK
4.24l;   // The 'l' must be part of the literal
4.24L;   // The 'L' must be part of the literal

4.2e4;   // OK
4.2e+4;  // OK
4.2e-4;  // OK
4.2e4f;  // The 'f' must be part of the literal
4.2e+4f; // The 'f' must be part of the literal
4.2e-4f; // The 'f' must be part of the literal
4.2e4F;  // The 'F' must be part of the literal
4.2e+4F; // The 'F' must be part of the literal
4.2e-4F; // The 'F' must be part of the literal
4.2e4l;  // OK
4.2e+4l; // OK
4.2e-4l; // OK
4.2e4L;  // OK
4.2e+4L; // OK
4.2e-4L; // OK

4.2E4;   // OK
4.2E+4;  // OK
4.2E-4;  // OK
4.2E4f;  // The 'f' must be part of the literal
4.2E+4f; // The 'f' must be part of the literal
4.2E-4f; // The 'f' must be part of the literal
4.2E4F;  // The 'F' must be part of the literal
4.2E+4F; // The 'F' must be part of the literal
4.2E-4F; // The 'F' must be part of the literal
4.2E4l;  // OK
4.2E+4l; // OK
4.2E-4l; // OK
4.2E4L;  // OK
4.2E+4L; // OK
4.2E-4L; // OK

// character and c-string literals:
'a';            // OK
u'a';           // The 'u' must be part of the literal
U'a';           // The 'U' must be part of the literal
L'a';           // OK
"I am a Text";  // OK
u"I am a Text"; // The 'u' must be part of the literal
U"I am a Text"; // The 'U' must be part of the literal
L"I am a Text"; // OK

// also all raw string literals are not supported

// with C++17 the following literals will also become correct:
// --> begin C++17
u8'a';
u8"I am a Text";
// <-- end C++17

Comments (5)

  1. Daniel Houck

    Since the creation of this issue, u, u8, and U prefixed string literals were fixed, but not the character literals and not anything else.

    Would a PR for some of these issues be welcome? I haven’t made the changes so I can’t guarantee I’ll be able to fix them, but I plan to at least give it a shot and I don’t expect any issues.

    If you would like a PR, I do have a question. How should user-defined literal suffixes be handled? Should they be part of the literal like the builtin suffixes are, or should they be some token type under Name (or just Name itself, as they are now)?

  2. Log in to comment