N2998
Literal Suffixes for size_t

1. Changelog

1.1. Revision 1 - June 17^th, 2022

• Add constraint to prevent size_t literals from becoming extended integer types (will be a proper Constraint Violation).

1.2. Revision 0 - April 12^th, 2022

• Initial Release 🎉

• Targeting C2y/C3a (after C23).

2. Design

This is a (minor) compatibility issue with literal suffixes that may be potentially used in shared code in the future. C++ adopted suffixed literals for size_t in [p0330]. The design is as follows:

#include <stddef.h>

size_t ulit_val = 0zu;

The u, as with current literals, can be placed on either the left or the right of the z suffix to make it unsigned. For symmetry with existing suffix literal rules, it also has a signed variant. That type is the "signed integer type of size_t", which normally resolves to ptrdiff_t:

#include <stddef.h>

ptrdiff_t lit_val = 0z;

The signed variant lacks the u as a piece of the z suffix. This also matches the way printf adjusts specific codes to display size_t or ptrdiff_t-sized variables. The design is simple and, thankfully, compatible with C++. It also provides a way to avoid signed comparison warnings in compilers which implement more strict comparisons checks, e.g., when comparing a size_t value against some fixed constant value.

2.1. Even After Compatibility, Do We Really Need This?

Yes. The trip to put this paper into C++ was an extremely long one and came with a ton of reasons. All of it can be seen in [p0330] and 90% of that reasoning applies to C, especially in the face of _Generic.

3. Wording

Wording is relative to [N2912].

3.1. Intent

The goal of this wording is to provide:

• a literal suffix for both the signed (ptrdiff_t) or unsigned (size_t) variant of the type of the expression "sizeof(0)"; and,

• use the suffix z in either the form uz or zu to produce a size_t.

3.2. Specification

3.2.1. Add two new grammar productions to §6.4.4.1 Integer constants¸ Syntax, ¶1

6.4.4.1 Integer constants

Syntax

…

integer-suffix:

unsigned-suffix long-suffix_opt

unsigned-suffix long-long-suffix

unsigned-suffix size-suffix

long-suffix unsigned-suffix_opt

long-long-suffix unsigned-suffix_opt

size-suffix unsigned-suffix_opt

…

long-long-suffix: one of

ll LL

size-suffix: one of

z Z

3.2.2. Add two new table rows to §6.4.4.1 Integer constants¸ Semantics, ¶6; Modify ¶7

The type of an integer constant is the first of the corresponding list in which its value can be represented.

Suffix	Decimal Constant	Octal, Binary, or Hexadecimal Constant
…	…	…
z or Z	the corresponding signed integer type of the expression sizeof(0)'s type (6.5.3.4)	the corresponding signed integer type of the expression sizeof(0)'s type
Both U or U and z or Z	the same type as the expression sizeof(0)	the same type as the expression sizeof(0)

If an integer constant cannot be represented by any type in its list, it may have an extended integer type, if the extended integer type can represent its value. If all of the types in the list for the constant are signed, the extended integer type shall be signed. If all of the types in the list for the constant are unsigned, the extended integer type shall be unsigned. If the list contains both signed and unsigned types, the extended integer type may be signed or unsigned. If an integer constant cannot be represented by any type in its list and has no extended integer type, then the integer constant has no type. Additionally, if an integer constant with the suffixes z, uz, zu, Z, UZ, or ZU cannot be represented by any type in its list, then the integer constant has no type.

Forward references: preprocessing numbers (6.4.8), numeric conversion functions (7.22.1) , The sizeof and _Alignof operators (6.5.3.4) .